- Roundtable - muug.ca

mysql crash and non-restart due to RAM?
by Trevor Cordes 04 Jan '16

04 Jan '16

I had a box with 32GB RAM have its mysql just die (no logs, just die). init tried to restart it (RHEL6) and I got this: 160104 8:55:55 InnoDB: Initializing buffer pool, size = 3.9G 160104 8:55:55 InnoDB: Error: cannot allocate 4194320384 bytes of InnoDB: memory with malloc! Total allocated memory InnoDB: by InnoDB 48401248 bytes. Operating system errno: 12 InnoDB: Check if you should increase the swap file or InnoDB: ulimits of your operating system. InnoDB: On FreeBSD check you have compiled the OS with InnoDB: a big enough maximum process size. InnoDB: Note that in most 32-bit computers the process InnoDB: memory space is limited to 2 GB or 4 GB. InnoDB: We keep retrying the allocation for 60 seconds... The setting of a 4G buffer pool was working for months until just now. I reduced it to 3G and mysql restarted ok. Top shows: Mem: 32827576k total, 2274940k used, 30552636k free, 165052k buffers Swap: 2097148k total, 276068k used, 1821080k free, 413640k cached Cacti (up until mysql crashed, wink) shows ps load going up, but that is normal for this box, and RAM usage stayed constant, right before the crash. However, we did launch a long ps that spawns a lot of little ps's (like maybe 100 at a time). This is all normal for this box, and it never has done this before. I'm guessing RAM fragmentation: #cat /proc/buddyinfo Node 0, zone DMA 2 1 1 1 1 0 1 0 1 1 3 Node 0, zone DMA32 1815 1789 1553 1294 1007 688 415 212 103 67 477 Node 0, zone Normal 32859 67280 77301 62304 45390 29754 19144 11069 4786 530 177 (anything else to check?) Looks like tons of pages avail of every size, assuming mysql uses "normal". Any ideas? My only thought now is to reboot server more often to clear fragmentation. Maybe newer kernels solve this "bug" a bit better, but I'm stuck on RHEL6 right now.

2 4

UI convention to show collapsible content (+ and - icons vs left/right and down arrows)
by Fortress Software Inc. 04 Jan '16

04 Jan '16

(Adam, Trevor, and MUUG round table, I have added you to this discussion for you valuable input.) My kids and I have been discussing proper icons to distinguish collapsible content on the web (specifically on one of my new mobile friendly/responsive web pages: http://matchomatics.com/index.php?country=canada&page=why-matchomatics) Our discussions were for which icons identify the collapsible content behavior the best (arrows/chevrons vs +/-). After much more web based research I think I will stick to '<' for collapsed and 'v' for expanded as using +/- icons can confuse users that this is as an 'add/remove' action. And sorry Alex, I see only a small sampling of reference to an '^' arrow to show collapsed content. Interesting to note that when using arrows the icon shows the 'current' state but when using +/- the icons show the 'action' state. Don't you just love UI transitions, as in the 'hamburger' menu icon... Bradford C. Vokey - Director ============================== Fortress Software Incorporated ============================== 390 Provencher Blvd Unit F Winnipeg, Manitoba, Canada R2H 0H1 ----------------------------------------------- Phone: (204) 231-3969 or 1-800-665-0396 FAX: 1-800-513-9356 -----------------------------------------------

4 3

Bug doh moment of the week
by Trevor Cordes 31 Dec '15

31 Dec '15

Doh! Ya, that might be a problem... Date: Wed, 30 Dec 2015 17:51:54 +0000 (UTC) From: updates(a)fedoraproject.org Subject: [SECURITY] Fedora 22 Update: cups-filters-1.4.0-1.fc22 [ 1 ] Bug #1291227 - CVE-2015-8560 cups-filters: foomatic-rip did not consider semicolon as illegal shell escape character https://bugzilla.redhat.com/show_bug.cgi?id=1291227

1 0

Re: [RndTbl] Roundtable Digest, Vol 132, Issue 9
by Michael Doob 23 Dec '15

23 Dec '15

This exact situation has happened to me due to the following scenario: I rsync parts of my system nightly onto an external esata drive mounted in the file system. If something hiccups and the drive gets unmounted, the backup is done onto the root file system at the mount point, and then when the drive gets remounted the files on the root system are hidden in just the way just as you described you described. The script now looks at what is mounted on the file system before rsync runs. It's obvious once you see it (just like mathematics). Cheers, Michael On 15-12-23 12:00 PM, roundtable-request(a)muug.mb.ca wrote: > Send Roundtable mailing list submissions to > roundtable(a)muug.mb.ca > > To subscribe or unsubscribe via the World Wide Web, visit > http://www.muug.mb.ca/mailman/listinfo/roundtable > or, via email, send a message with subject or body 'help' to > roundtable-request(a)muug.mb.ca > > You can reach the person managing the list at > roundtable-owner(a)muug.mb.ca > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Roundtable digest..." > > > Today's Topics: > > 1. "Missing" disk space (Adam Thompson) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Tue, 22 Dec 2015 12:02:05 -0600 > From: Adam Thompson <athompson(a)avant.ca> > To: roundtable(a)muug.mb.ca > Subject: [RndTbl] "Missing" disk space > Message-ID: > <CAFGv8b9jCKjwVhRjJLWm7cSWz9GoxQ5ZYO+AE_gCRsU0rhP8Vg(a)mail.gmail.com> > Content-Type: text/plain; charset="utf-8" > > Documenting for posterity, and in case anyone else runs into the same > problem: > > > I ran into a situation today where df(1) showed me a filesystem that was > getting pretty full (109GB used out of 146GB), but when I used du(1) to > find the space hogs, it was only able to locate <30GB of data on the > partition. > > (Hint: it was the root partition.) > > The culprit turned out to be 87GB of files "hidden" underneath > mountpoints. If you create /mnt/xyz, copy files into /mnt/xyz/*, then > mount another filesystem (local or remote) at /mnt/xyz, the original files > are still there, still taking up disk space, but unavailable. > > Luckily, this was a Linux system, and I was able to get access to those > files using a rebind mount: > - mkdir /tmp/xyz" > - "mount -o rebind /dev/mapper/root_lv /tmp/xyz" > - "cd /tmp/xyz/mnt/xyz" > - "du -sh *" > ...yup, there's 87GB of files that I *don't* see under /mnt/xyz. > > It's pretty obvious that the files were intended to live on a remote > fileshare, but got copied in at a point in time when the remote filesystem > wasn't mounted. Now I'm mv(1)'ing 87GB of data back across to a Windows > server across a CIFS mount... time to go for lunch, I guess! > > [image: Avant logo] *Adam Thompson* > Senior Systems Administrator > *voice:* 204.789.9596 x24 | *email:* athompson(a)avant.ca | *web:* avant.ca >

1 0

"Missing" disk space
by Adam Thompson 22 Dec '15

22 Dec '15

Documenting for posterity, and in case anyone else runs into the same problem: I ran into a situation today where df(1) showed me a filesystem that was getting pretty full (109GB used out of 146GB), but when I used du(1) to find the space hogs, it was only able to locate <30GB of data on the partition. (Hint: it was the root partition.) The culprit turned out to be 87GB of files "hidden" underneath mountpoints. If you create /mnt/xyz, copy files into /mnt/xyz/*, then mount another filesystem (local or remote) at /mnt/xyz, the original files are still there, still taking up disk space, but unavailable. Luckily, this was a Linux system, and I was able to get access to those files using a rebind mount: - mkdir /tmp/xyz" - "mount -o rebind /dev/mapper/root_lv /tmp/xyz" - "cd /tmp/xyz/mnt/xyz" - "du -sh *" ...yup, there's 87GB of files that I *don't* see under /mnt/xyz. It's pretty obvious that the files were intended to live on a remote fileshare, but got copied in at a point in time when the remote filesystem wasn't mounted. Now I'm mv(1)'ing 87GB of data back across to a Windows server across a CIFS mount... time to go for lunch, I guess! [image: Avant logo] *Adam Thompson* Senior Systems Administrator *voice:* 204.789.9596 x24 | *email:* athompson(a)avant.ca | *web:* avant.ca

1 0

OpenCSW Solaris packages
by Adam Thompson 18 Dec '15

18 Dec '15

https://www.opencsw.org/

3 9

Map of NSA spying on Canadian data
by Adam Thompson 18 Dec '15

18 Dec '15

http://www.theregister.co.uk/2015/12/17/canadian_live_route_map_shows_nsa_s… Neat... -Adam

1 0

math puzzle with U of M solution
by Trevor Cordes 09 Dec '15

09 Dec '15

As discussed at the meeting last night. Here's a fun math problem where the solution was submitted by a U of M student to Mike 20 years ago. Not often Winnipeg gets mentioned in international press! And the solution is even more nuts than I said last night: you also have to use imaginary numbers! Linux Mag makes you pay $2.95 paypal to read the whole article (you can read the first bit free). Worth it if you're interested. I think if you wait a few months they release the article for free. http://www.linux-magazine.com/Issues/2015/179/Perl-Math-Problems

1 0

FS: power supply lots
by Trevor Cordes 09 Dec '15

09 Dec '15

Forsale: 1999-2004 era used computer power supply lots. Seeing if anyone might have any interest, for a project, spares, parts, etc. They are older style (pre-24pin era). These are all standard form factor (PS2, fits all standard size MATX and ATX cases), 20-pin ATX connector (not the 24-pin required today). All PS's are used, most are in working condition, a few might have bad caps or dying fans (details are available). If anyone is interested I can run more thorough tests, I have a testing tool. Price would be quite inexpensive per unit, open to reasonable offers. Will separate if desired. LOT 1 (the newest): qty 7; with auxillary 4pin (2x2) square connector ("P4" connector for extra CPU power introduced in the P4 era). LOT 2: qty 20 or so; without aux 4pin; from P2/P3 era. (Also have AT power supplies if everyone ever needs any.) Email me separately, thanks!

1 0

setting "word-chars" in gnome-terminal
by Trevor Cordes 02 Dec '15

02 Dec '15

As per a comment I made maybe a year ago at a general meeting during roundtable regarding gnome-terminal removing the ability to set your own set of "word stop characters". This annoyed me (as an admin and programmer) to no end because all of a sudden I was forced to accept that double-clicking a word would also select : and #, and some other crazy characters that aren't "word chars" to me. They finally fixed this bug earlier in the year. Fedora 22 has the fix (vte 0.28 and gnome-terminal 3.16 and newer in any OS). However, the fix does not provide a GUI way to change the chars. Furthermore, the new method is pretty confusing. I haven't seen a good explanation/documentation yet on the net, in the bugzillas or elsewhere. Here's mine: First, get your gnome-terminal (g-t) "profile id". Right-click in your g-t -> Profiles -> Profile Preferences (make sure your desired / usual profile is selected). The UUID-ish looking thing in the General tab after "Profile ID" is what you need. You can select it, and paste it below (or retype it). For ease of manipulation, we'll put that profile-id into a shell var: csh/tcsh: set z=b1dcc9dd-5262-4d8d-a863-c897e6d979b9 sh/bash/ksh: z=b1dcc9dd-5262-4d8d-a863-c897e6d979b9 Note, that string might be the same for you, that's my default profile id, and it seems to pop up in other people's comments so it may be some kind of hardcoded value for the default profile. See what your values are set to: dconf read /org/gnome/terminal/legacy/profiles:/:$z/word-char-exceptions dconf read /org/gnome/terminal/legacy/profiles:/:$z/word-chars Your word-char-exceptions may be blank, and your word-chars may be set to something if you are on a box you've upgraded from old versions. Anyhow, you'll notice word-chars is being ignored. It's obsolete. Hose it (or just ignore it, but it confused me to no end until I confirmed with a dev it was being ignored): dconf reset /org/gnome/terminal/legacy/profiles:/:$z/word-chars Now we set the word-char-exceptions we want. Here's the confusing part: this value is what we WILL consider a word-char, and so chars you put in here will be selected when you double-click. Even worse, A-Za-z0-9 seem hardcoded in there, so you don't need to specify them. The nomenclature stinks because to me what you are setting aren't "exceptions", they are actual word-chars. Perhaps it should have been named word-char-additions. Here's how you set it: dconf write /org/gnome/terminal/legacy/profiles:/:$z/word-char-exceptions '@ms "-/(a)_&.?"' If you had word-chars previously set and liked it, just remove A-Za-z0-9 from that list and put the remainder chars in the word-char-exceptions list. You must move the dash to the front, right after the double-quote, like in a regex character class. You can specify ranges (in ascii) of chars using the dash elsewhere. I have no idea what the @ms and extra quoting is for, but the dev put that in his example and it works. I'm not sure how you'll get a ' or " in the list, but I'm sure it's possible (single quote should be easy but who knows how dconf wants the double escaped). As a programmer, making either quote a word char is insanity so I leave that as an exercise for the reader. The write to dconf takes effect immediately in your running g-t's, no need to restart anything. I (strongly) assume this writes to a persistent store somewhere so you don't have to do it on each boot. The devs say they'll put the GUI option back in the preferences dialog sometime in the future. That'll be much nicer! Why it was ever removed is beyond me. Now, onto another bz as to why the --title option in g-t was removed... Big thanks to Egmont Koblinger for being (one of) the driving force behind getting this option put back in.

1 0