| clort | hi | 01:17 |
|---|---|---|
| bru | Hello, clort | 01:33 |
| clort | what are you working on bru? | 01:45 |
| systemdlete | Alert to MATE desktop users: There is a SLOW memory leak in mate-panel in the clock applet. I am finding that it is eating about 132 bytes per minute (approx.). I was having strange freezeups, and subsequent log analysis showed me that the system was running out of memory. I used my handy memory leak program to hone in on it. This has been reported to MATE devs as issue 809 (see https://github.com/mate-desktop/mate-panel/ | 03:46 |
| systemdlete | issues/809) | 03:46 |
| systemdlete | Disabling the "show seconds" will alleviate this. MATE devs are considering eliminating the option. Just FYI, in case you are experiencing strange freezeups with MATE 1.20.4 | 03:47 |
| man_in_shack | memory leak. in clock | 03:49 |
| man_in_shack | wat | 03:49 |
| systemdlete | man_in_shack: See link above. | 03:51 |
| man_in_shack | i know | 03:51 |
| man_in_shack | it's more like, how did they even manage that? | 03:51 |
| systemdlete | (you said wat) | 03:51 |
| systemdlete | oh | 03:51 |
| man_in_shack | crazy stuff | 03:51 |
| systemdlete | their solution is to remove that feature. | 03:51 |
| systemdlete | effective, but... | 03:52 |
| systemdlete | If a baby is crying, do you toss it in the trash? | 03:52 |
| man_in_shack | won't that just slow the memory leak to 132 bytes per hour? (: | 03:52 |
| man_in_shack | systemdlete: yes. babies are gross | 03:52 |
| systemdlete | ? | 03:52 |
| systemdlete | Supposedly, if one disables the "show seconds" the problem goes away. However, upon further inspection here, I am finding that might not be the cure. | 03:53 |
| systemdlete | On one system with MATE, disabling seconds does seem to do the trick. But on the other, nearly ideentical system, it doesn't. | 03:55 |
| systemdlete | Well, it is not the clock applet this time. It's the mate-panel itself. | 03:56 |
| systemdlete | crud... it is happening on both after all. Previously, with seconds enabled, it was ticking 132 bytes per minute. Now it seems to have slowed down, but it is still eating and not releasing memory. | 03:59 |
| fsmithred | at 1/60 of the previous rate? | 04:02 |
| systemdlete | no | 04:02 |
| systemdlete | maybe 1/2 | 04:02 |
| systemdlete | or so... | 04:02 |
| fsmithred | you running someting other than beowulf? | 04:03 |
| systemdlete | nope. Beowulf on both VMs. Actually, star linux, but same thing, right? | 04:05 |
| systemdlete | On one system, the latest "eat" came 20 minutes after the previous one. Probably around the time I disabled the "show seconds" but I also deleted the netspeed applet and since that reading, I have also deleted the eyeballs applet. | 04:06 |
| systemdlete | The only stuff in the panels now are the menu, parcellite, volume slider, UPS meter, and weather clock. | 04:07 |
| systemdlete | and system monitor. | 04:07 |
| systemdlete | At any rate, something seems to be sloppy, memory management-wise | 04:09 |
| systemdlete | fsmithred: I tend to leave my VMs running. I reboot them as needed, and always when there is a vbox update. | 04:10 |
| Xenguy | Sheesh, just fix it | 04:10 |
| Xenguy | Make it work and not leak memory | 04:10 |
| systemdlete | Xenguy: Agree. | 04:10 |
| Xenguy | Chop chop said the user, haha | 04:11 |
| systemdlete | Chop chop said the butcher, to his programmers | 04:14 |
| Xenguy | systemdlete: You code? | 04:14 |
| systemdlete | a little, yeah. | 04:14 |
| Xenguy | Cool, I learn what I can | 04:15 |
| systemdlete | But normally, the solution is not to eliminate things. Usually, one first determines the source of the problem and see if it can be corrected. | 04:15 |
| Xenguy | The root problem(s), or the vision | 04:23 |
| systemdlete | fsmithred: I would say that now, after observing for 1/2 hour, the consumption is about once every 20 or 30 minutes (same 132 byte increase). | 04:29 |
| systemdlete | I notice that marco seems to be getting fat. Really fat. | 04:29 |
| Xenguy | Huh, it would be nice if they didn't bloat MATE | 04:36 |
| Xenguy | Up to know it seems fine, but lean is always better for me | 04:36 |
| Xenguy | er, up to now | 04:36 |
| systemdlete | In the past hour or so I've been monitoring my systems, one VM shows me that marco has grown by 518K. It just jumped by 150K, yet I had not opened or changed any windows. Not sure why it would suddenly grab memory. | 04:42 |
| Xenguy | huh | 04:49 |
| clort | how do i see which key code in /usr/share/X11/xkb/symbols corresponds to a certain key on keyboard | 05:27 |
| clort | grr | 05:27 |
| clort | xev no helping | 05:27 |
| clort | maybe things like AE10 AD07 are defined in keyboard geometry | 05:28 |
| I_am_nobody777 | mason: Hi , sorry , Missed you again . I resolved it. Turns out I didn't enable contrib repo . I searched for iucode in pkginfo page but it returned nothing before , searching iucode* brings it up and it is in contrib repo . I successfully installed intel-microcode . Thanks . | 06:23 |
| I_am_nobody777 | fsmithred: Hi , No problem .I understand it's experimental. I was just saying , I don't bother to use su - and typing toor . Also , copy the readme files to devuan user too , it would be useful . The memory test utility throws an error " error : version too old for 32bit boot. " , My laptop have Insyde h20 bios and the "EFI revision" is 2.40 according to rEFInd . I don't use memory tests but you know others may use . I am not complaining , I am | 06:35 |
| I_am_nobody777 | trying to be helpful . Thanks for your great work . | 06:35 |
| I_am_nobody777 | what's the normal utc time everyone gathers ? It would be nice to engage in conversation . | 06:37 |
| gnu_srs | My solution to the memory leak in MATE is to regularly kill the mate-panel process, it respawns. top with the M key reveals how many % memory is eaten by mate-panel. | 12:24 |
| clort | well that's unfortunate | 12:39 |
| * enyc meows | 19:22 | |
| * MinceR meows | 19:22 | |
| * DPA meows | 19:34 | |
| * debdog chases cats | 19:38 | |
| * MinceR chases debdog and goes for the throat | 19:42 | |
| * debdog behaves unimpressed | 19:43 | |
| * onefang tosses all the animals into #devuan-offtopic. | 19:50 | |
| * debdog hides from golinux (no hard feelings, just kidding) | 19:52 | |
| debdog | mastercat | 19:52 |
| systemdlete | firefox-esr is leaking. | 22:44 |
| systemdlete | sorry, I mean mate-panel is leaking, but it is a very slow leak. I think firefox is leaking a bit, but it has also given some memory back too. Whereas mate-panel just continues to grow, albeit slowly. | 22:46 |
| systemdlete | Disabling the "show seconds" feature of the clock applet has helped -- a lot. But still it grows. | 22:48 |
| rwp | systemdlete, I'm curious now... How fast? What's the rate of growth? | 23:31 |
| systemdlete | 132 bytes per interval. The interval got longer when I disabled the "show seconds" So now it is only growing about 3x per hour. | 23:33 |
| systemdlete | But get this, rwp: I am told that mate 1.20 is EOL for 2 years and that I need to upgrade to a more modern distro. | 23:34 |
| systemdlete | (mate 1.24 is the official now) | 23:34 |
| systemdlete | rwp: Before disabling "show seconds," it was growing about once a minute or two. | 23:34 |
| systemdlete | I'm running beowulf here... | 23:35 |
| systemdlete | (sorry for this scatological response... still on my 2nd cup, and it's already early afternoon) | 23:35 |
| clort | offtopic | 23:36 |
| systemdlete | sorry clort. I honestly thought memory leaks were a support issue. | 23:37 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!