--- Log opened dom dic 25 00:00:05 2022 --- Log opened dom dic 25 11:05:22 2022 11:05 -!- Irssi: #friendica: Total of 11 nicks [0 ops, 0 halfops, 0 voices, 11 normal] 11:05 -!- Irssi: Join to #friendica was synced in 18 secs --- Log opened dom dic 25 11:20:16 2022 11:20 -!- Irssi: #friendica: Total of 11 nicks [0 ops, 0 halfops, 0 voices, 11 normal] 11:20 -!- Irssi: Join to #friendica was synced in 18 secs 12:58 < fikabot> πŸ’¬ Running out of memory? Why? 12:59 < fikabot> πŸ’¬ Betreff: Cron /usr/bin/php bin/worker.php 12:59 < fikabot> πŸ’¬ 12:59 < fikabot> πŸ’¬ Fatal error: Allowed memory size of 1073741824 bytes exhausted (tried to allocate 4096 bytes) in /public_html/src/Database/Database.php on line 574 12:59 < fikabot> πŸ’¬ this is from the cron-job 13:00 < fikabot> πŸ’¬ php-fpm7.4 -v 13:00 < fikabot> πŸ’¬ PHP 7.4.33 (fpm-fcgi) (built: Nov 8 2022 11:40:37) 13:01 < fikabot> πŸ’¬ Debian bullseye and apache2 --- Log opened dom dic 25 17:26:49 2022 17:26 -!- Irssi: #friendica: Total of 11 nicks [0 ops, 0 halfops, 0 voices, 11 normal] 17:27 -!- Irssi: Join to #friendica was synced in 18 secs 17:40 < vladimir> hi 17:40 < vladimir> problem --- Log opened dom dic 25 17:45:17 2022 17:45 -!- Irssi: #friendica: Total of 12 nicks [0 ops, 0 halfops, 0 voices, 12 normal] 17:45 -!- Irssi: Join to #friendica was synced in 16 secs 17:47 < vladimir> pumpio addon behaves weird 17:51 < fikabot> πŸ’¬ > <@opendoor:matrix.org> Betreff: Cron /usr/bin/php bin/worker.php 17:51 < vladimir> https://friendicarg.nsupdate.info/display/69172192-1063-a85d-e451-326640085910 17:51 < fikabot> πŸ’¬ > 17:51 < fikabot> πŸ’¬ > Fatal error: Allowed memory size of 1073741824 bytes exhausted (tried to allocate 4096 bytes) in /public_html/src/Database/Database.php on line 574 17:51 < fikabot> πŸ’¬ > 17:52 < fikabot> πŸ’¬ 17:52 < fikabot> πŸ’¬ What's your per process memory limit in your fpm conf? Friendica log or server log showing errors? 17:53 < vladimir> https://friendicarg.nsupdate.info/log/friendica.log 18:02 < vladimir> https://reisub.nsupdate.info/filehost/files/RCq.png 18:02 < vladimir> https://reisub.nsupdate.info/filehost/files/7hX.png 18:12 < fikabot> πŸ’¬ > <@jbc:matrix.jb-net.us> What's your per process memory limit in your fpm conf? Friendica log or server log showing errors? 18:13 < fikabot> πŸ’¬ 18:13 < fikabot> πŸ’¬ You could see this in my posted error mail: 1024MB 18:23 < vladimir> memory_limit = 830M 18:36 < vladimir> up 1500mb 18:37 < vladimir> up pm.max_children = 35 18:38 < vladimir> I look forward to this new setup. improve performance 19:38 < fikabot> πŸ’¬ > <@fikabot:matrix.org> πŸ’¬ (vladimir) [libera] https://reisub.nsupdate.info/filehost/files/RCq.png 19:38 < fikabot> πŸ’¬ 19:39 < fikabot> πŸ’¬ Looks like you got a lot of gab.best with various subdomains trying to ping you. Probably another DDOS style attack like the activity pub troll did a few weeks ago. I'd block *.gab.best at the node level or even your firewall. 19:50 < fikabot> πŸ’¬ > <@opendoor:matrix.org> Betreff: Cron /usr/bin/php bin/worker.php 19:50 < fikabot> πŸ’¬ > 19:50 < fikabot> πŸ’¬ > Fatal error: Allowed memory size of 1073741824 bytes exhausted (tried to allocate 4096 bytes) in /public_html/src/Database/Database.php on line 574 19:50 < fikabot> πŸ’¬ > 19:50 < fikabot> πŸ’¬ 19:50 < fikabot> πŸ’¬ Depending on the size of the node you'd want to dedicate quite a bit of memory to innodb_buffer_pool_size in your mariadb conf file. 19:51 < fikabot> πŸ’¬ If you run out of RAM to cache your DB and it has to go to disk, it'll get waaay slower. 19:54 < fikabot> πŸ’¬ > <@jbc:matrix.jb-net.us> If you run out of RAM to cache your DB and it has to go to disk, it'll get waaay slower. 19:54 < fikabot> πŸ’¬ 19:54 < fikabot> πŸ’¬ Thanks. What is a "good" size for innodb_buffer_pool_size? I do not think the machine will run out-of-memory, because its 8GB for this virtualhost. 19:56 < fikabot> πŸ’¬ I'd say at least 50% of your usual free memory, if not a little more. So on your host if its usually at 6 GB free out of 8 GB, I'd give innodb pool size another 3 GB at least. 19:57 < fikabot> πŸ’¬ The more the better, but not so much your run out of total memory. πŸ™‚ 20:02 < fikabot> πŸ’¬ I will give this a try. 20:03 < fikabot> πŸ’¬ 20:03 < fikabot> πŸ’¬ https://mariadb.com/kb/en/innodb-system-variables/#innodb_buffer_pool_size 20:03 < fikabot> πŸ’¬ The primary value to adjust on a database server with entirely/primarily InnoDB tables, can be set up to 80% of the total memory in these environments. See the InnoDB Buffer Pool for more on setting this variable, and also Setting InnoDB Buffer Pool Size Dynamically if doing so dynamically. 20:04 < fikabot> πŸ’¬ Yep, in my experience it's the primary value I've noticed that can affect Friendica performance. 20:44 < fikabot> πŸ’¬ > <@jbc:matrix.jb-net.us> Yep, in my experience it's the primary value I've noticed that can affect Friendica performance. 20:44 < fikabot> πŸ’¬ 20:45 < fikabot> πŸ’¬ I have now tried this: https://raw.githubusercontent.com/major/MySQLTuner-perl/master/mysqltuner.pl 20:45 < fikabot> πŸ’¬ (from friendica-help: Skripte wie tuning-primer.sh und mysqltuner.pl, die den Datenbankserver analysieren und Hinweise darauf geben, welche Werte verΓ€ndert werden kΓΆnnten.) 20:45 < fikabot> πŸ’¬ 20:45 < fikabot> πŸ’¬ Result: 20:45 < fikabot> πŸ’¬ innodb_buffer_pool_size (>= 6.0G) if possible. 20:45 < fikabot> πŸ’¬ innodb_log_file_size should be (=768M) if possible, so InnoDB total log files size equals 25% of buffer pool size. 20:48 < fikabot> πŸ’¬ Mysqltuner is a good guide but it's good to sanity check it too. I'd just try increasing pool size 2-3GB and see how it performs. The tuner can base values off of current DB size, so if you haven't optimized your DB in a while, it may overshoot. 20:50 < fikabot> πŸ’¬ And obviously its recommendation would max out our close to max out your physical RAM, which has its own problems. 20:53 < fikabot> πŸ’¬ > <@jbc:matrix.jb-net.us> And obviously its recommendation would max out our close to max out your physical RAM, which has its own problems. 20:53 < fikabot> πŸ’¬ 20:53 < fikabot> πŸ’¬ These are also my thoughts, but give it a try and then: 20:54 < fikabot> πŸ’¬ 20:54 < fikabot> πŸ’¬ Variables to adjust: 20:54 < fikabot> πŸ’¬ *** MySQL's maximum memory usage is dangerously high *** 20:54 < fikabot> πŸ’¬ *** Add RAM before increasing MySQL buffer variables *** 20:56 < fikabot> πŸ’¬ Yeah, fortunately MariaDB can recognize that at least! So may have to try smaller increments to see when you start running into the max warnings. Any little bit you increase ought to help. 20:58 < fikabot> πŸ’¬ > <@jbc:matrix.jb-net.us> Yeah, fortunately MariaDB can recognize that at least! So may have to try smaller increments to see when you start running into the max warnings. Any little bit you increase ought to help. 20:58 < fikabot> πŸ’¬ 20:58 < fikabot> πŸ’¬ 4GB was too much also, but 3GB seems ok for the script. 21:00 < fikabot> πŸ’¬ So far so good! I'd just keep an eye on total RAM usuage and MariaDB usage too. With any sort of increase you ought to notice a pretty noticeable increase in performance and less in delivery queues. πŸ‘ --- Log closed lun dic 26 00:00:04 2022