wiki server failing - postgres error

If you are running Server 4, things have changed a lot.

sudo serveradmin status postgres

... is no longer supported there, but the error message is garbage.

Logs? Backups?

Move to the Trash, but don't empty:

/Library/Server/Wiki

Restore the above folder from a backup that predates the error.

Reboot and empty the Trash.

I have relocated my services data to an external drive so should I remove the wiki folder from here and replace it

Yes. Only that. You should also be aware that moving the data to a secondary volume is the best way to trigger many of the bugs in Server.

Hi, Linc. Thank you so much for all of your attention to these support communities. I have made use of many of your solutions in the past and am hopeful you will come through for us once again.

I will add what I can to the discussion, though I am little more than a "hobby" level administrator. So, please forgive my ignorance in advance.

1) Running OS X 10.10.4 and Server 1.4.3 - this point update was installed about 1 week before the wiki went down.

2) Even prior to the point update to the Server app, users were seeing wiki instability with exception messages appearing in place of pages. It was one of the motivations for moving forward with the update. I had researched the error in the past and never found a solution, but the problem would resolve with a system restart so it didn't seem any more severe than many other hiccups which have sorted themselves out through updates over the years. This link reports the same error we were seeing: http://serverfault.com/questions/698520/os-x-yosemite-server-wiki-stops-working - though I can't say for sure this is related to the present service failure.

3) I have been troubleshooting this issue since around July 25th when the service went down overnight and would not return the following morning. As with the instability issue above, the Server app continued to report that is was On and Available - all green lights - but only blank pages appeared in place of the earlier exception messages. Other services appeared to be functioning - though we only use caching, file sharing, time machine, DNS, and Open Directory. Attempted the following:

a) restarting the wiki service in the server app

b) system restart

c) server app replacement (leaving data in place)

- This resulted in an indefinite hanging (twice over night) of the Server app installation process when it reached "Updating Wiki Service." In troubleshooting this issue from community forum information, I believed the problem was the Postgres service - based on things like the original poster described above. However, if Grant Bennet's comment above is true, then that may have been an erroneous conclusion as I was reliant upon the same Command Line tools

d) OS reinstall (leaving data in place) followed by server app replacement (leaving data in place)

- Same result as c.

e) Wiki folder replacement from time machine backup

- Same result as c.

f) Complete removal of Server app as you detail here: Uninstall OS X Server completely in Yosemite (including full /Library/Server folder)

- This resulted in a successful Server app installation and working wiki services - though empty, without any of our wiki data

g) Various attempts to either "migrate" or "restore" the wiki data from time machine backups

- sudo wikiadmin migrate -r [Path to Data] :: I could not get this to work, as the former "Collaboration" folder no longer exists, and attempts to direct this process to anything related to the Yosemite wiki file data structure failed to import any data (Error: "Repository does not contain a Groups, Users, or User Preferences folder, so not recognized as a 10.6 Wiki repository.") Does this procudure no longer work in Yosemite?

- sudo psql -U collab -d postgres -f [Path to Dump] -h /Library/Server/Wiki/PostgresSocket AND

- sudo pg_restore -h "/Library/Server/Wiki/PostgresSocket" -d collab -U collab --single-transaction "/Path/to/location/of/collab.pgdump" :: Neither of these methods would work because I lacked the relevant dump file and could no longer create such a file without successfully loading the wiki services

h) Restore entire /Library/Server folder from time machine backup and replace Server app

- Began with restoration points from just prior to failure but encountered the same indefinite hanging of the Server app installation at the Wiki setup stage

- PERHAPS MOST SURPRISINGLY, When I fell back prior to even the installation of the latest point update (which had continued to function for weeks after this backup) the installation process again stalled at the wiki setup. I believed this would succeed because it predated the failure by weeks and had even functioned (though with the periodic exception thrown, as mentioned earlier) serviceably for several weeks after. Perhaps I am missing an important step here.

There are other steps I have taken which did not seem as relevant, so that is the summation of my experience with the issue. I recognize that I may have made many mistakes along the way, but I believed it would be recoverable because I had time machine backups to fall back on. Perhaps that confidence was misplaced? I am now concerned that I may never recover this important resource for my company.

Any insight would be very much appreciated.

*Server is 4.1.3, not 1.4.3 - transposed numbers there 🙂

The wiki service in Yosemite server suddenly stopped working with the error:

Caught exception "Connection to DB failed" [CSDatabaseError] executing route /app-context/wiki/

I of course started with turning wiki service on and off a couple times, with no luck. I also turned off the checkbox to enable WebDAV access to wiki files, since we don't need that anyway. Still wasn't working. So I dug into the postgresql database the wiki accesses, to see if I could figure out anything. I didn't see anything obvious in the log files, it just appeared that it was trying to back up while the database system was shutting down. However, the postgres instance owned by _teamsserver was definitely not running.

I happened to notice the pg_hba.conf file (/Library/Server/Wiki/Database.xpg/Cluster.pg/pg_hba.conf) was larger than it should be, up to almost 200K lines. The line

local replication all trust

was repeated many times, and at the end was a lot of blank lines. I deleted all the extra lines, and then cleaned up the stale /Library/Server/Wiki/PostgresSocket/.xpg_ctl.pid file as well, and then the database started up just fine. Note: it doesn't actually start up until you turn wiki service back on and access the wiki.

I'm not sure how the pg_hba.conf file ended up in that state, or even whether it was the problem or the stale .xpg_ctl.pid file. Hopefully this will help somebody else fix their wiki service someday.

I am having the same problem.

And fix mentioned by Linc didn`t worked.

attached screenshot with error. displayed on webpage and in log

Same problem all of a sudden. Answer by jcrate fixed the issue for me. But I have to repeat the procedure once a week, it seems.

(What's going on Apple? DNS settings GUI in Server app (4.1.5, Yosemite 10.10.5) also failing all of a sudden. Something in the latest updates must be rotten).