=== JanC_ is now known as JanC [08:13] Good morning === frickler_ is now known as frickler [09:22] question - /proc/sys/kernel/core_pattern is set to /tmp/core.%e.%p.%t, * soft core 500000 & root hard core 1000000 in limits.conf, the application segfaults and no corefile is found anywhere [09:22] ubuntu server 16.04 x64 [09:23] so, how do I go about enabling core dump files ? === ashleyd is now known as ashd [10:04] Hello all [10:04] what is the fast way to backup/recover a mysql database using MyISAM engine ? [10:04] I've found mysqldump to be very slow [10:04] with myisam thats kinda all you go [10:04] *got [10:05] with innodb you can do incremental backups but myisam is not transactional so your only choice is a full table dump [10:06] ddellav, I see thanks for your your answer [10:07] you're welcome Genk1 [11:31] also depends on database size how fast it is of course... [12:34] and what filesystem your using [12:35] personally, I had always used lvm snapshots to backup mysql myisam [12:35] but it has been awhile, and use innodb for just about everything now [12:51] patdk-lap: even LVM2 snapshots are not that good for consistent MyISAM state, but they usually work. [12:52] Genk1: you should really ask in #MySQL or #MariaDB for faster alternatives to 'mysqldump'. But that's really the right way to do it. If it is slow that may mean that your MyISAM files need compacting [12:53] depends on the definition of "slow" and considering database size [12:54] meanwhile, like Walex said, using fs snapshots to back up the database is illadvised as the database has its own regime of syncing states with the underlying fs, so you might be snapshotting incomplete states. [12:55] postgres has the ability to do WAL archiving. I don't know if that's possible with MySQL. Also, consider using real time replication for backup. [12:55] eh, if it is an atomic snapshot, it is no more unsafe than handling an unscheduled power loss [12:56] correct, but "unschedules power loss" is not a "consistent backup state" that you want your data BACK UP to be in :) [12:56] *d [12:57] and iirc, "unscheduled power loss" with MyISAM tables leads to "lots of corruption and pain". :) [12:57] well, then i'd certainly not use that for any data i'd like to keep. :) [12:58] if dumps are inadequate, replication is really the best, consistent form of back up. [13:00] dumps were hard for us, so we setup replication, then tuned that to the point that we could get dumps made on the replica so we had both [13:36] hmm, the database doesn't have it's own syning time, that is what a flush command is for [13:36] now if the filesystem supports flushing also, would be requred for lvm snapshot to work [13:36] but if every level is flushed, it's a perfect usable method [13:37] though personally, using a mysql slave database, and doing your backups on there, is better, if your concerned about downtime [13:48] depending on the dependencies, you could possibly dump some tables separately too... [13:56] patdk-wk: the files are mmaped and there's no guarantee that at every given moment, data in files represents 100% the in-memory state. [14:01] blackflow, did the default change? it was not using mmap by default normally === skylite_ is now known as skylite [14:44] patdk-wk: afaik it mmaps since 5.1. but now that you mention it, I could be wrong, yes. [14:45] I see support for mmap in 5.1 [14:45] but it is not enabled, except for packed myisam files, and those are read-only [17:04] hi [17:04] can someone help me with a routing problem? [17:07] anyone in here? [17:07] kinda [17:07] haha [17:08] is there a way to force a different subnet gateway to work instead of using the route command or post up? [17:12] anyone [17:12] ? [17:12] what's a subnet gateway? [17:14] show an example. Explain what it's doing, what the actual results are, what you'd want it to do instead [17:15] eg. pastebin current outputs of "ip a" and "ip r" and your network config (/etc/network/interfaces, or whatever else you use) [17:15] Also, don't say "instead of using (some perfectly reasonable options)" without explaining why you propose not to use them [17:17] well am going to try [17:17] the ip and gateway are on different subnet [17:18] so if i put gateway on the /etc/network/interfaces it can not be bring up [17:18] now if i remove it and add route add default gw (gw ip) [17:18] as a command it works [17:19] now i have to through cloud-init add default gateway to the /etc/network/interfaces due to automation [17:20] so basically a fix could be adding a command before the network starts to delete the line gateway xxx.xxx.xxx.xxx from /etc/network/interfaces before bringing it up on boot [17:21] This sounds a bit scary and broken [17:21] i know [17:22] thats why am trying to find a better solution [17:22] since ubuntu doesnt allow a different subnet gateway on /etc/network/interfaces [17:23] I think you will need to paste the output of "ip route" if we are to have any hope of understanding your network setup [17:23] ok am gonna show with pics [17:27] maxb here is not working [17:27] https://i.imgur.com/z91hKJR.png === JanC is now known as Guest83509 === JanC_ is now known as JanC [17:29] hold on [17:30] This is not a sensible configuration [17:31] You can't use an address outside the subnet as a default gateway unless there's some other special routing going on [17:37] hi guys, after a dist-upgrade process my server is unbootable, Im getting this on the screen, any advice? http://imgur.com/a/l1hvV [17:38] robert45: press enter and examine? [17:39] for a (n entirely blind) guess: zfs related [17:40] tomreyn I appreciate the quick reply, Im now inside the shell. I did a service networking status and this is what I get, ifconfig shows correct network setup: http://imgur.com/a/bY7yz [17:41] robert45: you should start examingn the cause of the first error, not the last [17:43] robert45: on the screen shot you posted, the first error (which has not scroll off screen, i.e. there may have been others before this one) is about systems-tmpfiles-setup.service [17:43] the temporary / volatile file storage service did not start up properly. [17:44] it is suggested that you should run "systemctl status systems-tmpfiles-setup.service" to examine this issue. [17:44] note how the message right before this is about starting ZFS support via FUSE [17:45] tomreyn tx, Im running that command now [17:46] also note that ZFS-fuse is considered deprecated [17:47] use the native module instead [17:49] Hi! I want to cache HTTPS site contents such as Facebook Images and etc. I have added the following https_port 3126 tproxy ssl-bump generate-host-certificates=on dynamic_cert_mem_cache_size=4MB cert=/etc/squid3/lib/ssl_db/certs/myCA.pem connection-auth=off. This gives me errors. Is squid can not be run this way? [17:49] Please if anyone know the answer, PM me. [17:52] you might want to discuss the errors you get there. [17:53] in logs? [17:53] It just crashes. Segfault at xx.xx.xx [17:54] Can squid be run with Tproxy target for HTTPS traffic? [17:54] \ [17:54] Can you pls confim me that ^ [17:54] * tomreyn lacks Squid experience, can't tell [17:54] MASM: Can squid be run with Tproxy target for HTTPS traffic? [17:56] All I can tell is that caching SSL content contradicts with the idea of end-to-end encryption, it is only possible by creating a man-in-the-middle. Some companies and ... ISPs in some countries ... still do it anyways. [17:57] i don't know men, sorry... [17:59] http://wiki.squid-cache.org/Features/HTTPS [17:59] i assume "tproxy" stands for 'transparent proxy'. [18:03] tomreyn sorry I had some troubles with the systemctl command, this is what I get: http://imgur.com/a/x7p6l [18:04] I assume its because the filesystem is mounted as readonly [18:05] robert45: same here [18:05] Im rebooting it again to see if get why its not mounting properly [18:05] robert45: i would check syslog, go back to when you booted, then look through those lines until you hit the first error. [18:06] or start with "dmesg -T | less" - sometimes this is faster [18:07] tomreyn tx, doing that now [18:07] one (of many) possible explanation might be file system corruption on /etc or / (i do not know your partitioning scheme) [18:08] tomreyn ugh! I hope not, its a production server :( [18:09] tomreyn ok I caught the first error, I dont think it could be related, do you? http://imgur.com/a/p0OXS [18:09] apparmor failing could be related, yes [18:10] tomreyn well, its the same error, systemctl shows a read-only filesystem [18:10] you surely have current backups, and if it's an important system you surely considered a HA configuration? [18:10] when systemctl apparmor thing [18:11] so check dmesg + syslog, find out why it's r/o [18:31] tomreyn tx, I was able to bring it online, few service still failing though [18:34] tomreyn for some reason its mounting /dev/sda1 as ro, but if I use mount -o remount,tw it mounts it ok, any ideas? [18:36] tomreyn nevermind, I was able to fix it, it was all fstab related [18:36] its showing 16.04 instead of 16.10, not sure it didnt upgrade all the way through [18:38] robert45: so you upgraded from 16.04 to 16.10 recently, or tried to? appartenly it didn't happen if lsb_release -sd still shows 16.04 [18:39] tomreyn I upgraded from 14.04, it shows 16.04 now [18:39] first time [18:39] its showing 16.04 instead of 16.10, not sure it didnt upgrade all the way through [18:39] this is contradictionary [18:40] tomreyn not sure if I understood. I was looking to upgrade to 16.10 [18:40] tomreyn so I upgraded from 14.04, do I need to run another upgrade? [18:40] either you last upgraded from 14.04 to 16.04, or from 16.04 to 16.10 [18:41] only one of them can be correct [18:41] tomreyn I upgraded from 14.04 to 16.04 [18:41] Im looking to upgrade to the latest version 16.10, is it possible? [18:42] yes, personally i prefer to run LTS releases on servers, whoever, do not like to have to upgrade every 6 months [18:42] s/whoever/however/ [18:42] also LTS releases tend to be a lot more stable === ejat_ is now known as ejat [18:43] tomreyn I see. I ran do-release-upgrade and got "No new release found" [18:43] tomreyn oh I just noticed 16.10 its not a LTS release [18:44] tomreyn so you dont recommend to upgrade from LTS to latest non LTS ? [18:44] correct, 16.10 is not LTS. and personally i prefer to only run LTS on servers. in even more so in your case. ;) [18:44] correct, 16.10 is not LTS. and personally i prefer to only run LTS on servers. and even more so in your case. ;) [18:44] ^typo fixed [18:45] first of all, i would suggest you ensure your previous upgrade succeeded, and the system is in a good condition. [18:46] !lts | robert45 [18:46] robert45: LTS means Long Term Support. Until 12.04 LTS versions of Ubuntu were supported for 3 years on the desktop, and 5 years on the server; since 12.04 (Precise Pangolin) LTS versions will be supported for 5 years on the desktop and server. The latest LTS version of Ubuntu is !Xenial (Xenial Xerus 16.04.1) [18:46] tomreyn yes I think I wont upgrade it, Im scary to suffer any other downtime like this one [18:46] tomreyn understood, so far everything looks good! :) [18:47] also be very sure of the non-lts support lifetime [18:48] it used to be 18months, but now is like 9 I think, or maybe 6 [18:48] and system upgrades mean downtime [18:49] robert45: if you like i can check your output of "apt-get update", "apt-cache policy", "apt-get -V -f install", "apt-get -V -y --simulate dist-upgrade" later. [18:49] or worse, a broken package during upgrade [18:49] robert45: .. post them to a pastebin if so, e.g. using !pastebinit [18:49] !pastebinit | robert45 [18:49] robert45: pastebinit is the command-line equivalent of !pastebin - Command output, or other text can be redirected to pastebinit, which then reports an URL containing the output - To use pastebinit, install the « pastebinit » package from a package manager - Simple usage: command | pastebinit -b http://paste.ubuntu.com [18:50] * tomreyn taking a shower now [18:50] sure, running those now [18:53] tomreyn patdk-lap looks like some errors indeed: http://paste.ubuntu.com/23701740/ [18:53] how should I proceed? [18:56] how can i run post-up command outside of /etc/network/interfaces? [18:58] erick3k: cut-n-paste to the shell? [18:59] andol i meant to run on startup outside interfaces, [18:59] in shell says command not found [19:01] erick3k maybe put link in bin path [19:03] erick3k, maybe create a shortcuts in local/bin [19:04] MASM can you link me to an example and a little newbie [19:05] erick3k, shortcuts : http://unix.stackexchange.com/questions/226315/how-to-use-ln-s-to-create-a-command-line-shortcut [19:05] erick3k, program executable everywhere : http://unix.stackexchange.com/questions/3809/how-can-i-make-a-program-executable-from-everywhere [19:06] MASM am not sure that would work i want post-up route add xxx.xxx.xxx.xxx dev eth0 to run outside /etc/network/interfaces file [19:07] just that command [19:07] because /interfaces will be modified by cloud-init [19:08] If cloud-init is incorrectly changing your config, then fix that, rather than trying to paper over it elsewhere [19:09] huh if i knew how [19:09] i would [19:09] is it possible to run that command outside interface or not? [19:10] like on centos [19:10] you just add route-eth0 file [19:17] Ubuntu's network configuration chooses to give you one readable file rather than littering bits and pieces across multiple files like CentOS [19:20] robert45: i don't see errors there. I see a notice, "N: Ignoring file '50unattended-upgrades.ucf-dist' in directory '/etc/apt/apt.conf.d/' as it has an invalid filename extension", and i see a lot of no longer needed (due to changed dependencies) packages which can likely be removed (using apt-get --purge autoremove) [19:21] maxb so only solution is to have a custom script or have cloud-init run it? [19:21] tomreyn ok do you recommend to --purge autoremove or should I kept those for just inc ase? [19:21] yes [19:22] If you're using cloud-init to configure your networking, it needs to do it *right* [19:23] If it isn't, you need to work on addressing that [19:24] robert45: yes maybe dont purge immediately [19:24] there's zfs-fuse in there [19:24] got it, maxb what about if i make a file executable here /etc/network/if-up/ with that command? [19:25] would that work? [19:40] tomreyn sorry got dc'ed, thanks so much for your help, wish you a happy new year! === JasonMc92 is now known as CodeMouse92 [22:15] maxb you still here? [22:16] remember about the network? [22:16] found a solution :)