Community
Participate
Working Groups
1. Log in to my OrionHub account 2. Try to load my workpace -- can't > {"HttpCode":500,"Message":"Error saving state","Severity":"Error","Code":0} 3. Try to create a new site configuration -- can't > {"HttpCode":500,"Message":"Error saving state","Severity":"Error","Code":0} This is an account I just created a few minutes ago.
Trying to update my user profile got me this: > "{"HttpCode":500,"DetailedMessage":"No space left on device","Message":"Can not update user: XXXXXXXXXXXXX@gmail.com","Severity":"Error","Code":4}"
It looks like we have about a 20GB partition. The current workspace with all user data is 5.3GB. I'll poke around and see if there is a user taking lots of space. Webmaster, any chance this partition could be increased? We can certainly do some cleanup but I can imagine this easily requiring more when there is a good size user base. We currently use 1.5GB on our internal self-hosting server for a team of about 10 people.
There is only one user that stands out as far as disk usage: matthieu dartiguenave is using 500MB, but it at looks like legitimate content (git clones). I've reclaimed a couple of GB by clearing up some old content, but there is another problem: some of the metadata files are corrupt, probably because the server attempted to write to them and failed, leaving an empty file.
Here's the contents of the admin dir: 775M current 1.8G eclipse-orion-0.2RC3 4.6G eclipse-orion-0.3M1 4.0K finddos.pl 34M m7-old 68K navigate-table.js 0 nohup.out 1.4G old 4.0K perfmon 28K perfmon.log 5.3G serverworkspace.bug.364496 4.0K start_orion 4.0K sysmon.pl 4.0K uptime_log 4.0K userlist 4.0K wipe_workspace.sh Surely, there is some redundant junk that can disappear? 1.4G old, 1.8G 0.2RC3, m7-old. And what's that 5.3G directory?
(In reply to comment #4) > Surely, there is some redundant junk that can disappear? 1.4G old, 1.8G > 0.2RC3, m7-old. And what's that 5.3G directory? Working on it... Those old directories are backups of user data. I'm assuming the data on this machine isn't being backed up in any other way. I've deleted some of the backups to free up space, but I'm stepping carefully to make sure I don't lose any more user data. The 5.3G directory (serverworkspace.bug.364496), is the user data at the moment we ran out of disk space. I moved it rather than delete it so I can take the time to dig into how things failed, and there might be data in there that users would like recovered. Anyway, still working on cleaning things up and I should be able to knock it down another 5GB. A bit of extra space would definitely help though... my television has more disk space than this web server ;)
(In reply to comment #5) > my > television has more disk space than this web server ;) Yes, but can you run Orion on it? Nevermind -- you probably can.
Entered bug 364507 about improving metadata file writing in orion to avoid data loss when disk is full.
Cleaned up space. We should be ok for the immediate future.
