!fnetworks Tentatively planning some downtime for the SUR1 (home of the Nu and N0 GS instances) and USS (home of the N5 instance) servers. It should be intermittent and run from Friday 2021-01-01 through Sunday 2021-01-03.
I think y’all know I favor the #Fossil #SCM / #DVCS over #Git (though I’m giving #Mercurial ( #Hg ) another look as well. I’m resource-constrained as well as energy-limited, but I have been thinking of standing up a miniature software forge for !FNetworks projects and users, likely using Kallithea (which can use both Hg and Git) and also having some Fossil repos (but not as part of a forge).
Fortunately, we have not had a problem with this, but "loli" and similar content is not welcome on any !fnetworks server. When I get some time, I'll write a draft ToS that makes it clear.
@xrevan86 discovered a faster db query for filling timelines. At some point after this is in the repo, we'll be updating all !FNetworks instances to the newest code.
Adding more #php-fpm workers helped up to a point, but I think NFN (Nu) is beyond that point already.
I don't think we have anyone on a !Fnetworks instance that would do this, but just for reference, please refrain from recommending non-scientific "cures" for this deadly disease.
I recently turned AP on for N5. Will wait until I have tested updates and fixes before deploying it to Nu and N0.
(N1-N4 existed in the past. By now, there should be no servers that still have them in their databases, so they may return, assuming !fnetworks picks up enough users and an admin team to handle that)
If I get TME* this week, I'll set up a #Friendica instance for !fnetworks ... membership is only available to people with currently active Federati accounts.
I was going to set up #Pleroma, but the PleromaFE interface has too much #JavaScript / #JabbaShit piled up ... It causes my browsers to repeatedly pop up warnings about a hung script.
!fnetworks Outage of about an hour on the SUR1 server. Hosting company Ramnode investigated and fixed it, though the message they sent about the cause sounds fishy to me.
Considering reverting !fnetworks SUR1 server's #PHP version to 7.2 to try to fix the slowness issue. Right now, I'm restarting the daemon a couple of times per day. I thought it was a #MySQL query issue, but generally, restarting PHP speeds things up considerably.
Still trying to track down the slowness. Watching top tells me that most of the time for a request is spend by PHP-FPM. Will be trying
“Pay special attention to the slowlog settings (request_slowlog_timeout and slowlog directives), that is, through this log and a reasonable amount of timeout, you can easily see which php calls from your application take longer than expected and start debugging them right away.” — https://cwiki.apache.org/confluence/display/HTTPD/PHP-FPM [apache org]
Federati Networks is the name for the Federati.net and associated sites and the (hopefully soon-coming) organization founded to fund and manage the sites.