|
| Greg's diary |
| May 2026 |
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Friday, 1 May 2026 | Dereel | |
| Top of page | ||
| next day | ||
| last day |
|
More laptop setup fun
|
Topic: technology, opinion | Link here |
Spent what time I had today continuing with setting up bde.lemis.com. I wanted to use it in the kitchen, which of course required changing the network configuration: lagg. Once again I have instructions from Google Gemini. Basically it entails loading the lagg.ko module, creating a device and setting a configuration. The twins write:
# Bring the physical ports up without IPs
ifconfig_em0="up"
wlans_iwn0="wlan0"
ifconfig_wlan0="WPA"
# Create the virtual failover interface
cloned_interfaces="lagg0"
ifconfig_lagg0="laggproto failover laggport em0 laggport wlan0 DHCP"
Why DHCP? I know the address I want, so I set it. Start lagg. No more net. I cut off the branch I was sitting on. And since bde is dependent on the rest of the network for many files, it hung too.
Restart and set /etc/rc.conf.d/lagg:
cloned_interfaces="lagg0"
ifconfig_lagg0="laggproto failover laggport em0 laggport wlan0 192.109.197.144
And it still hung. Looking locally, I had:
em0: flags=1008843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST,LOWER_UP> metric 0 mtu 1500
options=800020<JUMBO_MTU,HWSTATS>
ether 6c:3b:e5:f3:fa:79
inet 192.109.197.144 netmask 0xffffff00 broadcast 192.109.197.255
media: Ethernet autoselect (1000baseT <full-duplex>)
status: active
nd6 options=29<PERFORMNUD,IFDISABLED,AUTO_LINKLOCAL>
wlan0: flags=1008843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST,LOWER_UP> metric 0 mtu 1500
options=0
ether 6c:3b:e5:f3:fa:79
hwaddr 24:77:03:d1:7c:28
inet 192.109.197.244 netmask 0xffffff00 broadcast 192.109.197.255
groups: wlan
ssid L3M15 channel 10 (2457 MHz 11g ht/40-) bssid f4:3e:61:48:6c:b2
regdomain FCC country US authmode OPEN privacy OFF txpower 30
bmiss 10 scanvalid 60 protmode CTS ampdulimit 64k ampdudensity 8
-amsdutx amsdurx shortgi -stbc -ldpc -uapsd wme roaming MANUAL
parent interface: iwn0
media: IEEE 802.11 Wireless Ethernet MCS mode 11ng
status: associated
nd6 options=29<PERFORMNUD,IFDISABLED,AUTO_LINKLOCAL>
lagg0: flags=1008843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST,LOWER_UP> metric 0 mtu 1500
options=0
ether 6c:3b:e5:f3:fa:79
hwaddr 00:00:00:00:00:00
inet 192.109.197.244 netmask 0xffffff00 broadcast 192.109.197.255
laggproto failover lagghash l2,l3,l4
laggport: em0 flags=5<MASTER,ACTIVE>
laggport: wlan0 flags=0<>
groups: lagg
media: Ethernet autoselect
status: active
nd6 options=29<PERFORMNUD,IFDISABLED,AUTO_LINKLOCAL>
Where did that 244 in the last address octet come from? It's in the DHCP range. To be investigated, along with the comment in the Gemini configuration that the IP address for the em0 interface should not be set. But there's time for that later.
|
Wiring bde for sound
|
Topic: technology, multimedia, opinion | Link here |
The other thing that I needed to do with bde was to get camera and microphone to work. I've established that the camera itself works, but can I get it to cooperate with a web browser? Tried it with Zoom, which confirmed that it could see me (in better resolution than pcwview), and it could play a sound better than what I got from Radio Swiss Classic yesterday. But no microphone input. How many causes can there be for that? Where do I start looking? I don't know, but I know something about when to start looking: not before tomorrow.
| Saturday, 2 May 2026 | Dereel | Images for 2 May 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
KL Hokkien Mee again
|
Topic: food and drink, opinion | Link here |
KL Hokkien Mee for breakfast today. It's been nearly two months since I last cooked it, and I had to check some details. Strangely, I missed the broth that goes into the noodles, though the rest of the sauces and cornflour were as normal. And strangely I didn't miss the broth. It didn't even seem too thick. What is right?
|
More fun with bde?
|
Topic: technology, opinion | Link here |
Cooking breakfast involved bde.lemis.com for the recipe. Nothing world-shattering, but maybe that was related to the reason why I missed the broth in the KL Hokkien Mee. But it brought a couple of things home: FreeBSD still doesn't install all normal fonts needed for X, so things like 猪油渣 render incorrectly. Once again I'm going to have to use the Linux fonts.
And the other thing? I hate laptop keyboards. Yes, they're an order of magnitude better than mobile phone surfaces, but that's not saying very much.
| Sunday, 3 May 2026 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Remote watering
|
Topic: animals, technology, opinion | Link here |
Chris Bahlo is currently in South Australia at some competition, so Yvonne is checking on her horses for her. Today I went with her, because she had reported that her Borzoi Fjodor had injured his paw.
Fjodor is now coming on 12 years old, quite an age for a Borzoi—we had never had one that lived much beyond 9 years of age. But they generally die of bone-related diseases, and Fjodor has spent most of his life in a dog run, where he doesn't get to run around very much. His paw injury was very minor, and it didn't need treating.
But of course we went around and looked at the horses, and discovered that one watering trough was completely dry, and the others not much better. There are no ball valves. How do we fill them? Where is the pump? Last time I didn't exactly crown myself in glory with my investigations, and Chris ended up with an unnecessary bill. So Yvonne contacted her, and... she turned on the pump, remotely! I wonder what connectivity she uses for that. Probably a mobile phone app, but what else?
|
Better cheese in Australia?
|
Topic: food and drink, opinion | Link here |
Finding cheese for fondue de fromage is always a problem. We have plenty in the freezer, but it's always worth looking for alternatives. And on asking for suppliers of Appenzeller in Melbourne, Google Gemini came up no fewer than 7 suppliers, including Alpine Express, located in Brisbane. But they send cheese nationwide at prices that I can accept, so they're worth considering.
In addition, they supply not one, but two different kinds of Appenzeller: “Rahmkäse - White Label and “Appenzeller Classic `Mildly Spicy'”. What does that mean? The twins tell me “The German name for Appenzeller White Label is Appenzeller Mild-Würzig.” And in English, „Mild-würzig“ means “mildly spicy”. About all I can glean is that both cheeses are only 3 months old, not what I'm looking for. Where's « surchoix »? Where's „extra“?
We have time. I'll keep an eye on supplies.
| Monday, 4 May 2026 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Health care in times of pain
|
Topic: health, technology, opinion | Link here |
A couple of days ago I received an SMS from Health First telling me that I needed a vaccination.
Why? I had arranged the vaccination with my doctor months ago. But more to the point, why an SMS? They're so unmalleable and unreliable. Called up today. No problems, says Shenae, we won't inform you any more. How about email? No, sorry, we only send “text”.
Oh smart.
|
More bde stuff
|
Topic: technology, opinion | Link here |
On with trying to get bde.lemis.com to work correctly. First, why does the microphone not work? Again Google Gemini to my aid. There's a button on the keyboard, second from the right:
If it lights orange (like here), the microphone is disabled. Press it and it will go out or light white.
But it didn't. Disabled from BIOS? Reboot and take a look. Password, please. But I don't know any password, and I can't ask Bruce. The twins again:
For EliteBook models, HP Support can generate a SMC.bin file tailored to your laptop's UUID (Universally Unique Identifier). You will need to prove ownership of the device.
And how do I do that? I didn't buy the device, and I don't have any paperwork. Anyway, I can try. Call HP on 02 8278 1039 (there's also a 0300 number, but they're more expensive). Another bloody voice non-recognition system. What kind of device is it? A list, not including laptops. Ah, that's a notebook. Are you calling about the progesss of a repair? No. Sorry, missed that. Are you calling about the progesss of a repair? No. Sorry, missed that. Enter 0 or 1.
What's wrong with a system that can recognize “laptop”, but not “no” in response to a yes/no question? In any case, I was connected to Neha (why do these people all have such unusual names?). The first thing she wanted to know was whether she could call back on this number in case of disconnection. Unusual, but why not? Then she wanted to know model and serial number, of course. Where's the serial number? On the underside.
Nope:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
That looks very much like I need to dismantle the device to find the serial number! Never mind, says Neha. Click on the icon at the bottom. No icon, of course. She didn't even ask what operating system I was running. Log in. Hang. Ah, yes, of course, I had disconnected the network cable to turn the thing upside down. But while I was fixing that, the call got disconnected. And sure enough, she called back almost immediately. But not as immediately as the next call got disconnected. And that was the last I heard of her.
Somehow HP have serious issues somewhere. I can't be overly bothered.
A bit more searching brought me to dmidecode, which can access internal data. And it gave me the serial number: 5CB2480314, and also the information that the processor is an Intel Core i5 3320M. But that's not much help.
So what do I do now? Find a potential jumper that can reset the BIOS data, including data that I might need? Or just give up and use an external microphone? Found one that cost $5.35 including postage, so I bought that and postponed the sound until next week.
Then the question of remapping the keyboard. Created a keymap file /usr/share/vt/keymaps/grog.kbd and loaded it. How about that, it really does work, and though I “didn't change anything”, Alt was also remapped to its correct place. But somehow it doesn't work (yet) on reboot.
And lagg? Tried the invocation exactly as stated, and how about that, it came up! What I missed was the first part:
# Bring the physical ports up without IPs
ifconfig_em0="up"
wlans_iwn0="wlan0"
ifconfig_wlan0="WPA"
And the result was that neither em0 nor wlan0 had any IP address. That belonged to lagg0. Of course, that used DHCP to get the IP address, coincidentally 192.109.197.244 instead of 192.109.197.144. OK, no worries. Bring up lagg with an explicit IP address:
-ifconfig_lagg0="laggproto failover laggport em0 laggport wlan0 DHCP"
+ifconfig_lagg0="laggproto failover laggport em0 laggport wlan0 192.109.197.144"
And it still used DHCP! Why? More RTFM needed.
| Tuesday, 5 May 2026 | Dereel | Images for 5 May 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Instant noodles again
|
Topic: food and drink, opinion | Link here |
Not much hunger this morning, time for another of these horrible instant noodle dishes, this time Pancit Canton. In principle nothing new, though this time I took the opportunity to get rid of some celery that was lying around:
|
|||||||||||||||||||||||
Another 2 packages to go. I'll be glad when this stuff is gone.
|
Macro pain
|
Topic: photography, opinion | Link here |
Taking photos this morning gave me a chance to check a number of things. First, the photos in the kitchen were the first with the new flash unit. How do things look? Here a photo of the cooking area:
|
|||||||||||||||||||||||
That was taken at f/8 instead of previously f/5.6, and that was relatively correct. That 1 EV more light, more than the difference between 300 and 400 J would suggest.
And here entire kitchen area:
|
|||||||||||||||||||||||
That looks reasonably uniform, though the flash to the right (the new Godox SK400II) was noticeably brighter, as shown by the shadows of the strap of the Nikon FM2 on the bench. And I was able to take this photo at f/16 instead of previously at f/11.
So what do I do? Swap the flash units? Maybe. I can test by dropping the output of the new unit to 225 J.
And then I wanted to take photos of the buttons on the EliteBook. How hard can it be? The point was to show the light in the microphone button, so I couldn't use flash. Used the Leica 25 mm f/1.4 Summilux to get enough light. Tried with the Olympus E-PM2, and despite confirming focus, it came out completely out of focus.
OK, try again with the Olympus OM-D E-M1 Mark II. Yes, acceptable:
|
|||||||||||||||||||||||
Why so far away? In true old Leica tradition, the Summilux can't focus any closer. OK, that's what extension tubes are for. Put on a 10 mm tube, and... the thing wouldn't focus at all. Why? It's nothing but a pipe. I need to check, and that could take time, but it looks as if there may be some electronic misunderstanding, maybe the autofocus for the OM-1 Mark II.
OK, nothing for it. Get the Zuiko Digital 50 mm f/2 macro and take my photos. That worked, and as I observed 60 years ago, the apertures in this range don't reflect the light gathering ability: the Zuiko required 2/3 EV more exposure than the Summlux, when it should have been 1 EV. But finally I had my photo.
|
More photos with the Nikon FM2
|
Topic: technology, history, opinion | Link here |
I was recently reminded that I still have a film in the Nikon FM2. It has the same rating as my OM System OM-1 Mark II, and the flash sync speed is the same (1/250 s, extraordinary for a camera introduced in 1984), so I can use the same settings. But how do I record them? I took four images of the kitchen along with the ones above. I think they were:
| Pan | f/8 | |
| Kitchen | f/16 | |
| ? | f/16? | |
| Pan | f/11 |
But where should I collect this information? And in passing, focussing with this kind of camera is so much more difficult than with a modern camera. I need to compare it with the Pentax Spotmatic, which I found really good in those even further-off days.
|
bde lives!
|
Topic: technology, opinion | Link here |
I am gradually configuring bde.lemis.come the way I want it. All I need is a sane keyboard and to get lagg to work. Checked the details in various places. Yes, what I had written was correct, with one or two exceptions: I hadn't specified a net mask, and the end of the line was missing. It was on the keyboard map specification too.
OK, fix that, and... it works! Here the diff:
--- /etc/rc.conf.d/lagg 2026/05/05 02:54:47 1.1
+++ /etc/rc.conf.d/lagg 2026/05/05 04:04:00
@@ -5,4 +5,4 @@
ifconfig_wlan0="WPA"
# Create the virtual failover interface
cloned_interfaces="lagg0"
-ifconfig_lagg0="laggproto failover laggport em0 laggport wlan0 192.109.197.144"
\ No newline at end of file
+ifconfig_lagg0="laggproto failover laggport em0 laggport wlan0 192.109.197.144 netmask 255.255.255.0"
Now why is the system so slow?
| Wednesday, 6 May 2026 | Dereel | Images for 6 May 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Web server disk overflow
|
Topic: technology, opinion | Link here |
Every morning I check my external web servers for load and storage space: the log files can overflow quickly, and I still don't have automatic log rollover fixed. And today I had the tell-tale from mail.lemis.com (also one of the www.lemis.com): no mail for several hours.
Sure enough, yesterday and today:
=== grog@lax (/dev/pts/6) ~ 19 -> df
Filesystem 1048576-blocks Used Avail Capacity Mounted on
/dev/ufs/rootfs 54,550 35,762 14,423 71% /
devfs 0 0 0 100% /dev=== grog@lax (/dev/pts/6) ~ 20 -> df
Filesystem 1048576-blocks Used Avail Capacity Mounted on
/dev/ufs/rootfs 54,550 51,408 -1,222 102% /
devfs 0 0 0 100% /dev
Those values are in megabytes, so I had over 15 GB of log files overnight! What was it? An error message that I had programmed for exceptional circumstances. No long exceptional, and the message was long, 5489 characters, and not very intelligible:
[Tue May 05 22:51:54.673445 2026] [php7:notice] [pid 48076] [client 51.75.117.47:35852] assoc (/mythweb/tv/opensearch, /... /grog/email/), referer: https://www.google.com/search?hl=en&q=testing
Apart from the fact that it can fill up the log file quickly, I no longer know what it is trying to tell me. What I really need is a stack trace.
|
More breakfast experiments
|
Topic: food and drink, opinion | Link here |
I eat a lot of noodles for breakfast, and I have a whole freezer drawer full of precooked noodles. But more often than not I cook an individual noodle basket, so the drawer stays full to overflowing.
So today I tried an experiment designed mainly to use up scraps:
| quantity | ingredient | step | ||
| 150 g | Beijing noodles | |||
| 59 g | Pork | |||
| 10 g | Spring onion | |||
| 1 | Dried chili | |||
| 11 g | Celery | |||
| 57 g | Carrot | |||
| 28 g | Mushroom | |||
| 43 g | Lettuce | |||
| 10 g | Soya sauce | |||
| 10 g | Dark soya sauce |
The result:
|
|||||||||||||||||||||||
And it was boring! That's not the fault of the ingredients, more what is missing. And in passing the noodles were too soft. Maybe I should reduce the cooking time (from currently 7 minutes), but it seems that I have tried that before and it didn't work.
|
Autofocus evolution
|
Topic: history, photography, opinion | Link here |
Found a potentially interesting article on the evolution of autofocus today. I can see a couple of potential errors, but first I need to read it first.
|
Learning about Israel
|
Topic: history, opinion | Link here |
Read in the Times of Israel today: a free course in the history of Israel, from the Jewish perspective, by Rabbi Mike Uram. It starts out emphasizing neutrality in viewpoints, but somehow I wonder if that isn't just a more subtle way of stating their case. To be followed.
| Thursday, 7 May 2026 | Dereel | Images for 7 May 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Simplifying phat si-io
|
Topic: food and drink | Link here |
I rather like phat si-io, but it's difficult to get right. How do you minimize oil usage? How do you char the food enough? Do you need to char the food so much? Decided to try for a relatively “western” approach, using tagliatelle instead of kway teow, and not frying as powerfully.
Things didn't quite work out like that. The tagliatelle that I had frozen in the freezer were far too much, so in the end I used kway teow after all. But this time I didn't try to char them as much, which might have worked. But my experiments aren't over.
|
Still more bde fun
|
Topic: technology, opinion | Link here |
Things worked relatively well in the kitchen with bde, but the keyboard mapping was still wrong. Investigations showed that the CapsLock key still generated CapsLock, though xev claimed Control_L. More questions to Google Gemini, which came up with a new invocation that I didn't know:
setxkbmap -option ctrl:nocaps
That works, presumably only when the key has already been remapped to Control, but that was the case. Unfortunately, I had issued the command when CapsLock was on, so I couldn't turn it off again! But the twins had an answer for that, too:
xdotool key Caps_Lock
The only problem there is that I can't find xdotool, but I found a way by invoking setxkbmap without options. At least the keyboard is working now. Now how do I keep it that way? One issue remains: if I try to change to a vty, the screen goes blank and stays that way until reboot. So for the time being I'll leave it the way it is
|
More extension tube fun
|
Topic: photography | Link here |
I've had these extension tubes on my desk for the past couple of days, since I had difficulties with focusing. That was with the OM System OM-1 Mark II and the Leica 25 mm f/1.4 Summilux. Could one of them be the issue?
So I tried with completely different hardware: the Olympus OM-D E-M1 Mark II and the M.Zuiko Digital 45 mm f/1.8. And that worked fine.
OK, try with the Summilux on the E-M1. Yes, I got focus, but only very close, and it seems that it didn't try to focus anywhere else. Back to basics: the Summilux focuses down to 0.3 m. Rearranging the lens formula 1/u + 1/v = 1/f, I get 1/v = 1/f - 1/u.
Oh. That's much less of a difference than I expected. Use the extension tube to make 35 mm, v becomes 85 mm when the lens is set to ∞. And that could be the case.
I'm surprised. The implication is that extension tubes are pretty much useless for lenses of such short focal length. Looking back to the Good Old Days, a standard lens had a focal length of 50 or 58 mm, and the smallest tube (11 mm) would reduce the maximum focus of ∞ down to 28 cm. That leaves an gap between 28 cm and tye typical closest focus of 45 cm. And in well over half a century I hadn't noticed that. That's so surprising that I wonder if I have made a miscalculation, but it explains my observations perfectly.
|
Grilled Bratwurst
|
Topic: food and drink, opinion | Link here |
Bratwurst for dinner tonight. I wondered if it wouldn't be a good idea to grill them in an air fryer. Have I done this before?
Yes, many times, most recently only 5 months ago. And it didn't work quite as I had intended. Back to look at the YouTube clip that I had based that on, and followed further back. No, their method is not what I want. Grill them at 210° in the hair dryer.
How long? Let's observe. This time I did the obvious and put the sausages in a metal pan instead of on the grill, which would leave indentations. And I deliberately didn't slash them like I did last time.
The results? After 15 minutes at 210° they looked quite good:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
But Yvonne wanted more, and turning them over showed that some kind of cut would have been useful:
|
|||||||||||||||||||||||
Another 5 minutes like that and things were to Yvonne's satisfaction:
|
|||||||||||||||||||||||
So do it that way next time, after first slashing the skin.
| Friday, 8 May 2026 | Dereel | Images for 8 May 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Bacon and eggs again
|
Topic: history, food and drink, opinion | Link here |
I don't cook bacon and eggs any more. The last time was 8 May 1976, 50 years ago today. On that occasion (on a Saturday) I got a call from work and put it off. Two days later I was fired.
But for old times' sake (literally), today I cooked bacon and eggs again. The real issue is the eggs, not the bacon, though that came out somewhat uneven. From other experience, it's clear that the pan temperature is important. Measure it:
|
|||||||||||||||||||||||
110° sounds right. But it was still too hot (left egg):
|
|||||||||||||||||||||||
And in addition, and despite lots of oil, it stuck to this non-stick pan, causing it to fall apart when I removed it:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The result? One was alright, the other a bit of a mess:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
So in all likelihood that's the last time I ever cook bacon and eggs.
|
Multiple anniversary
|
Topic: history, Stones Road house, general | Link here |
Losing my job wasn't the only anniversary today, though it was the roundest. But 11 years ago we moved into our current house in Stones Road, Dereel, the longest we have ever lived anywhere. And there's more to come.
|
Pecking order
|
Topic: animals, opinion | Link here |
I processed some mince today, packed in one of Woolworths “we're environmentally friendly” single-use plastic trays, which was also slightly corrugated to retain some of the meat.
OK, give it to the cats. Mona first, but of course Bruno came along. Who would eat first? As it happened, neither: they happily finished off the rest together.
But then Larissa came along. Vicious hunting dog against two cats? Not quite:
|
|||||||||||||||||||||||
And she didn't dare go near the tray until the cats finished.
| Saturday, 9 May 2026 | Dereel | Images for 9 May 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Anniversaries, day 2
|
Topic: history, opinion | Link here |
Yesterday's anniversaries didn't come alone, though none were as round as the 50th anniversary. 51 years ago I married my first wife, Doris Pischke, and 29 years ago Yvonne, my second wife, arrived in Australia, marking the beginning of our residence here.
But wait, there's more! My daily calendar tells me:
May 9 The first parliament of Australia opens in Melbourne, 1901
May 9 VE day, end of Second World War, celebrated in many countries
VE Day was on 9 May? Not 8 May? Well, yes. The exact time was 23:01 on 8 May, Berlin (summer) time. That's 21:01 GMT, but more importantly 0:01 on 9 May in Moscow, or 7:01 on 9 May here. So 9 May it is.
|
Network tuning
|
Topic: technology | Link here |
I have a domestic Internet link providing 25 Mb/s down and 5 Mb/s up. By modern standards, that's slow, and I could easily update to something like 400/40. Why don't I?
One reason is the price, of course. 400/40 currently costs $119 a month. Even 100/20, the next step up, costs $93, and I currently pay only $75 on a legacy plan. I can afford that, of course—30 years ago I was paying much more for a 64 kB/s ISDN link, round $800 or so for the meagre amount of traffic I could afford.
But more to the point, it doesn't buy me anything. Most of my Internet traffic is file downloads, mainly video programmes. And those downloads typically max out at round 200 kB/s, or 1.6 Mb/s, far below my link speed. I can do them in parallel and get more, but why so slow?
TCP/IP tuning? A while back I investigated that, but for some reason I didn't implement it. Now we have Google Gemini. What does it tell me? Quite a lot, of course. It boils down to a number of sysctls:
sysctl kern.ipc.maxsockbuf=16777216
# Send/Receive Space:
sysctl net.inet.tcp.sendspace=262144
sysctl net.inet.tcp.recvspace=262144
# Auto-Scaling Limits:
sysctl net.inet.tcp.sendbuf_max=16777216
sysctl net.inet.tcp.recvbuf_max=16777216
And that increased the values:
kern.ipc.maxsockbuf: 2097152 -> 16777216
net.inet.tcp.sendspace: 32768 -> 262144
net.inet.tcp.recvspace: 65536 -> 262144
net.inet.tcp.sendbuf_max: 2097152 -> 16777216
net.inet.tcp.recvbuf_max: 2097152 -> 16777216
There's other stuff in there too which has long become default, like congestion control
=== root@teevee (/dev/pts/21) ~ 13 -> sysctl net.inet.tcp.sendbuf_auto
net.inet.tcp.sendbuf_auto: 1=== root@teevee (/dev/pts/21) ~ 14 -> sysctl net.inet.tcp.cc.algorithm
net.inet.tcp.cc.algorithm: cubic
And the results? Varied. Downloading from ARD increased to round 1 MB/s, but ZDF barely changed. That was half to be expected: downloads from SRF were frequently in the 1 MB/s range, so there's more to improving throughput than tuning the network stack. Still, a relatively quick improvement.
|
False hare
|
Topic: food and drink, language, photography | Link here |
Yvonne cooked „Falschen Hase“ for dinner this evening. What's that in English? I had guessed “false hare”, but Wikipedia just wants to call it meat loaf, ignoring the details. Maybe “mock hare” is the best suggestion I have seen.
For me it was mainly an opportunity to check the new flash relationships. This one, in front of the oven, looks correctly lit and exposed:
|
|||||||||||||||||||||||
But this one (f/6.3, where once f/5.6 was more appropriate) looks overexposed:
|
|||||||||||||||||||||||
And while this on the dining table is correctly exposed (f/5.6), it clearly shows the issues with a single flash there.
|
|||||||||||||||||||||||
What can I do there? I have a flash that I could theoretically put on the other side, but there's nowhere to put it. I think it's a case of “well don't do that, then”.
|
Why doesn't eBay send me mail?
|
Topic: technology, general, opinion | Link here |
I receive a lot of mail from eBay, and the email address I chose (not ebay@lemis.com) gets exposed to the sellers. So it's not surprising that I got spam and had to change the address.
That works well, but of course eBay is quite incompetent in many ways. One was sending me copies of messages I send to sellers. Tick “send me copy by email”, but nothing happens.
Today I sent a message to somebody and took a look.
May 9 00:03:58 lax postfix/qmgr[68036]: 48D3428754: from=<member@ebay.com>, size=52136, nrcpt=1 (queue active)
May 9 00:03:59 lax postfix/smtp[80565]: 48D3428754: to=<ebay@lemis.com>, relay=mx0.lemis.com[121.200.11.253]:25, delay=597, delays=596/0.03/0.59/0.23, dsn=5.1.1, status=bounced (host mx0.lemis.com[121.200.11.253] said: 550 5.1.1 <ebay@lemis.com>: Recipient address rejected: User unknown in local recipient table (in reply to RCPT TO command))
What was that? It's clearly from eBay, sent to the old address. The time looks right. I'll probably never know. I'll reinstate the address and see what comes through.
| Sunday, 10 May 2026 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Masur dal
|
Topic: food and drink, opinion | Link here |
I've been planning to cook this South African Biriyani for weeks now. But it's strange: it wants lentils. Specifically red lentils, masur dal.
How can that work? I've spent some time trying to work it out. Off to take a look at our reserves. Yes, red lentils that expired in December 2007. Too old? Never mind, another pack expired on 15 April 2015. Still too old? How about the split lenilts that expired on 11 June 2017? Google Gemini tells me that they're all far too old.
OK, we don't want to ruin the dish, so Yvonne bought a fourth package. And today I cooked them, for 10 minutes, more than enough. That's a big difference from other pulses, which can take hours. And for the fun of it, I cooked some of the newest of the old lentils. And they tasted pretty much the same. Tomorrow I'll try this recipe, originally from Gemini.
In passing, it's interesting to note how they swell up. The increase in weight isn't very different from noodles or rice, about 2.25× the weight, but the volume increases considerably.
|
More network throughput investigation
|
Topic: technology, multimedia, opinion | Link here |
So how much difference did yesterday's network tuning make? I had established that it made as good as no different to ZDF, but the throughput from ARD went up from round 200 kB/s to 1 MB/s. But there's more to network throughput than local tuning. What about the other end? What about the path in between? I had already established that SRF delivered that even before the tuning. And Australian sites such as ABC and SBS also worked better. What's it like now?
Better. Downloaded a series of episodes from SRF and got speeds between 2.82 and 2.90 MB/s, corresponding to 22.6 and 23.2 Mb/s raw data, about as much as can be expected from a 25 Mb/s link. In fact, watching with iftop showed that the raw throughput hit round 28 Mb/s; the throttling doesn't work completely, and since the link is capable of much more, I don't suffer from performance degradation during times of congestion.
And ABC? Also much better, but surprisingly less than SRF: betwen 2.57 and 2.64 MB/s (20.6 and 21.1 Mb/s).
So: there's every reason to believe that the current limiting factor is the speed of my link. I can upgrade, and I can expect even better performance. Should I? These are really batch operations, and downloading an episode is done in under 10 minutes. If the link speed doubles, I have it in under 5 minutes. Why should I pay money for that difference?
In passing, it's interesting to note the amount of data that I have transferred, more than ever before. So far this billing month (ending on Tuesday) I have downloaded 966 GB of data and uploaded 201 GB, a total of 1.167 TB. At the prices I was paying 35 years ago that would correspond to DM 525,150,000. In current terms, it corresponds to $ (AUD) 1,100,000,000. Of course, that could never have happened. At 9.6 kB/s the transfer would have taken 320 years.
How times change!
Only after writing the above did I discover that I've been there before, coincidentally exactly 5 years ago.
| Monday, 11 May 2026 | Dereel | Images for 11 May 2026 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Masur dal poha for breakfast
|
Topic: food and drink, opinion | Link here |
As planned, masur dal poha for breakfast today, or sort of. Instead of poha I had some left-over saffron rice. And once again I was bitten, badly, by the ambiguous units used. One onion? They're too polite to say. One teaspoon of turmeric powder? How much?
In the end, I used:
| quantity | ingredient | step | ||
| 7 g | mustard seeds | 1 | ||
| ghee | 1 | |||
| 7 g | cumminseed | 2 | ||
| 2 | fronds fresh curry leaves | 2 | ||
| 8 g | dried chilis | 2 | ||
| 90 g | Onion, finely diced | 3 | ||
| 16 g | garlic paste | 4 | ||
| 11 g | ginger paste | 4 | ||
| 4 g | turmeric | 5 | ||
| 400 g | Cooked red lentils | 6 | ||
| Garnish: Fresh lemon juice, coriander | 7 | |||
Fry the mustard seed in ghee until they pop.
Add cumminseed, curry leaves and dried chilis. Fry a little.
|
|||||||||||||||||||||||
Add the chopped onion and fry until translucent.
Add garlic and ginger and fry until hot.
|
|||||||||||||||||||||||
Add turmeric and fry a little.
|
|||||||||||||||||||||||
Add lentils and mix well.
|
|||||||||||||||||||||||
The recipe called for coriander (seed) powder, but I had none, so I left it out.
I then mixed about 100 g of this mixture with about 80 g of saffron rice:
|
|||||||||||||||||||||||
I didn't have any green chilis, so I used some chili leaves that I have had for ever:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The result:
|
|||||||||||||||||||||||
And how was it? Completely out of balance! It wasn't until I tasted it that I realized that there was no salt at all in the recipe. And it shows a Western habit of using far too little of all the spices (assuming that I guessed the quantities correctly). Next time I'll use maybe 10 g of salt and 2.5 times as much spice (to remain on the conservative side). The only thing that was too much was the chilis; I'll halve them.
|
SBS download fail
|
Topic: technology, multimedia, opinion | Link here |
After yesterday's success in faster downloading, how about trying SBS? I even found a programme that looked interesting.
=== grog@tiwi (/dev/pts/49) /spool/Series 15 -> yt https://www.sbs.com.au/ondemand/movie/the-fox/2380100675822
Downloading https://www.sbs.com.au/ondemand/movie/the-fox/2380100675822
youtube-dl -c -o %(title)s.%(ext)s --all-subs https://www.sbs.com.au/ondemand/movie/the-fox/2380100675822
[SBS] Extracting URL: https://www.sbs.com.au/ondemand/movie/the-fox/2380100675822
[SBS] 2380100675822: Downloading SMIL file
ERROR: [SBS] 2380100675822: Unable to download SMIL file: HTTP Error 410: Gone (caused by <HTTPError 410: Gone>)
What's that, apart from an incorrect status? The file is still there, and I can watch it via a browser if I can bear it. It seems that they have found yet another way to make life difficult for people.
Do I care? SBS has a strange collection of videos, and even the ones that do look interesting often prove to be not worth watching. But yes, I do care. There are browser plugins that claim to be able to do it, but that's not the way I like to do things. One more thing to investigate.
|
A microphone for bde
|
Topic: technology, multimedia, opinion | Link here |
My USB microphone for bde.lemis.com arrived today. Plug it in.
May 11 14:57:36 bde kernel: ugen0.2: <C-Media Electronics Inc. USB PnP Sound Device> at usbus0
May 11 14:57:36 bde kernel: uaudio0 on uhub0
May 11 14:57:36 bde kernel: uaudio0: <C-Media Electronics Inc. USB PnP Sound Device, class 0/0, rev 1.10/1.00, addr 1> on usbus0
May 11 14:57:36 bde kernel: uaudio0: No playback.
May 11 14:57:36 bde kernel: uaudio0: Record[0]: 48000 Hz, 1 ch, 16-bit S-LE PCM format, 2x4ms buffer. (selected)
May 11 14:57:36 bde kernel: uaudio0: Record[0]: 44100 Hz, 1 ch, 16-bit S-LE PCM format, 2x4ms buffer.
May 11 14:57:36 bde kernel: uaudio0: No MIDI sequencer.
May 11 14:57:36 bde kernel: pcm5 on uaudio0
May 11 14:57:36 bde kernel: uaudio0: HID volume keys found.
Test it. The real test was to work my way through the mess of the Zoom web site. And it didn't work.
Oh. That was potential microphone 1 out of no less than 8 devices. And it turned out that the real microphone was number 3. And it worked.
But then, just to annoy me, I ran out of swap space:
May 11 15:25:07 bde kernel: swap_pager: out of swap space
May 11 15:25:07 bde kernel: swp_pager_getswapspace(10): failed
May 11 15:25:08 bde kernel: swp_pager_getswapspace(20): failed
...
May 11 15:26:22 bde kernel: pid 17847 (firefox), jid 0, uid 1004, was killed: failed to reclaim memory
May 11 15:26:23 bde kernel: pid 29083 (firefox), jid 0, uid 1004, was killed: failed to reclaim memory
What went wrong there? Yes, with “only” 8 GB of RAM it's not the biggest, and it had “only” 4 GB of swap. But after firefox died, there was almost none in use, though a tell-tale 6 GB of RAM free. Maybe firefox complaining about Zoom?
That wasn't the end of it. X hung. The cursor jumped back and forth like I've seen in the past with x2x. Fortunately I was able to stop x2x on hydra:0.3, but the cursor continued to jump about on bde:0. Stopping and restarting the window manager didn't help. And I can't restart X on this machine, so I had to reboot.
And that showed me that my keyboard configuration is still not correct. This is so frustrating. I'll fix it tomorrow.
| Tuesday, 12 May 2026 | Dereel | Images for 12 May 2026 |
| Top of page | ||
| previous day |
|
Advertisement overload
|
Topic: technology, health, opinion | Link here |
Checking on the news this morning on albo.lemis.com, my mobile phone, I saw an advertisement telling me to ditch my relatively accurate blood pressure monitor and use a mobile phone app.
OK, I'll bite. How does that work? It even wants to measure blood sugar (but only in unaccustomed mg/dl) and BMI. Installed it, but I couldn't get it to work. Instead it was so full of advertisements that I couldn't find out what it was supposed to do, nor whether it could do it. Somehow these apps are getting so advertisement-oriented that they're completely useless. By the time I gave up, I had accidentally installed 5 different apps, all of which were overloaded with advertisements. One claimed, without reading, that my blood pressure was 119/79, and my pulse was 69. Is that just guesswork?
About the only thing of interest was one that claimed to read my blood pressure via the camera. Place your finger on the “camera”. Wait. Which camera? It turned on the light, which I thought was what I needed, but no, it was one of the other lenses. And at the end it came up with the decision that my pulse was 83/min.
And the blood pressure? Another dozen advertisements. And then, without any measurement, it decided that it was 100/75, and my pulse was 70. Was that the same app? By comparison, I measured 118/65 and a pulse of 87 (!) with a real blood pressure monitor.
So what's the point of these things? They're so inaccurate and hard to use that their suggestion of ditching the correct equipment is just plain stupid.
|
More bde fun
|
Topic: technology, multimedia, opinion | Link here |
Still more work on bde.lemis.com today. Yesterday I had established that the keyboard mapping still wasn't correct. Off to compare what I had done. The problem I had was that the key at bottom left (marked “ctrl” should be Meta. OK, check the keymap file (/home/local/X/keyboard/default-map). xev tells me that the key has the code 37:
KeyPress event, serial 33, synthetic NO, window 0xc00001,
root 0x3ec, subw 0x0, time 93545014, (77,123), root:(726,147),
state 0x0, keycode 37 (keysym 0xffe3, Control_L), same_screen YES,
And what's in /home/local/X/keyboard/default-map?
keycode 37 = Control_L NoSymbol Control_L
OK, fix that then. But while checking, the key generated Meta_L. Where did that come from? What am I missing?
What I did manage was to set up Nextcloud so that video and loudspeaker worked, the latter very quietly. It's too polite to test the microphone, which has multiple ways of describing itself. In the system probes I have
May 11 14:57:36 bde kernel: uaudio0: <C-Media Electronics Inc. USB PnP Sound Device, class 0/0, rev 1.10/1.00, addr 1> on usbus0
But that's not what the web browser wants. For some reason, despite my telling it to save the information, it kept coming back with microphone 1, the one that didn't work. And for reasons I really don't understand, it always comes up in laterally inverted mode. Why? I can correct it, but why should I? It seems to be some kind of brain damage in this area: Zoom does it too. But finally I was able to get something like a test done:
|
|||||||||||||||||||||||
Unfortunately, it's too polite to test my microphone. Fired up Zoom and confirmed that yes, it works. But while looking at that, looked at the memory usage:
last pid: 40873; load averages: 5.69, 2.75, 1.38; battery: 99% up 0+05:04:15 15:57:58
69 processes: 2 running, 67 sleeping
CPU: 95.0% user, 0.0% nice, 2.1% system, 1.0% interrupt, 1.9% idle
Mem: 2224M Active, 2187M Inact, 199M Laundry, 944M Wired, 596M Buf, 2227M Free
Swap: 4096M Total, 4096M Free
PID USERNAME THR PRI NICE SIZE RES STATE C TIME WCPU COMMAND
40577 grog 98 1 0 4197M 1237M CPU1 1 12:57 217.35% firefox
That looks normal enough. But while watching it, I saw
last pid: 40894; load averages: 6.30, 5.39, 3.30; battery: 99% up 0+05:11:27 16:05:10
PID USERNAME THR PRI NICE SIZE RES STATE C TIME WCPU COMMAND
40577 grog 94 3 0 13G 1633M select 3 31:22 254.01% firefox
That's a big increase for an idle system. Is Zoom to blame? Kill it (the tab):
last pid: 40895; load averages: 6.04, 5.40, 3.37; battery: 99% up 0+05:11:52 16:05:35
PID USERNAME THR PRI NICE SIZE RES STATE C TIME WCPU COMMAND
40577 grog 85 3 0 6050M 2735M select 2 32:35 334.97% firefox
last pid: 40895; load averages: 4.86, 5.21, 3.38; battery: 99% up 0+05:12:25 16:06:08
PID USERNAME THR PRI NICE SIZE RES STATE C TIME WCPU COMMAND
40577 grog 83 0 0 8353M 4737M select 3 34:05 27.29% firefox
The resident memory increased by a factor of nearly 4! Was that the reason for running out of memory yesterday?
|
Homemade chips: last gasp
|
Topic: food and drink, opinion | Link here |
I've been quite disappointed in the results of making chips myself rather than buying pre-prepared frozen chips. A year ago I bought a chip cutter:
|
|||||||||||||||||||||||
As the image shows, it's flimsy. I had great difficulties cutting the chips: the potatoes seemed to be firmer than the cutter. And the results were less than encouraging. I had gradually come to the conclusion that the thing was a waste of time and money.
But before finally discarding it, it occurred to me: these things cut two different thicknesses. The one I had been using was the finer of the two, roughly 8.5 mm square (64 chips max per cut). But there's a second cutter, offering roughly 11 mm mainly square (36 chips, some rectangular cross section). Try that?
The good news: the potato cut with no trouble at all. And the results didn't look that bad:
|
|||||||||||||||||||||||
But they were still so dark that I pulled them out of the fat 1 minute before they were due. And they weren't completely cooked. Clearly both the deep fryer and the chip cutter really wanted me to use this size.
But why? I don't like the look of the chips, they don't taste any better, and they're the wrong size for my taste. I won't throw the cutter away yet, but I don't see myself using it again in the foreseeable future.
In passing, weights again: 206 g of potato gave 117 g of chips.
Do you have a comment about something I have written? This is a diary, not a “blog”, and there is deliberately no provision for directly adding comments. It's also not a vehicle for third-party content. But I welcome feedback and try to reply to all messages I receive. See the diary overview for more details. If you do send me a message relating to something I have written, please indicate whether you'd prefer me not to mention your name. Otherwise I'll assume that it's OK to do so.
| Top of page | Previous month | Greg's home page | Today's diary entry | Greg's photos | Copyright information |
