Bugzilla – Bug 16285
Instability issues of SBS with WHS
Last modified: 2019-01-25 12:11:12 UTC
There is a critical issue with SBS on WHS. After restart it is inaccessible by controllers, players, and browser even if service started (this not a duplicate of the checkbox not sticking bug) and no apparent problem is in log files. Only fresh install works, and when when uninstalling SBS, under the Bin directory: mysqld.exe and faad.exe files are not removed. This prevents the installer from working. The files has to be renamed and than deleted. Note that these are with install and unintalls from WHS console. Only the manual deletion has to be done with RDP. I have been through this process about 10 times, it is consistent. I do not restart the WHS often, but it is annoying to go through this process each time. In order to reproduce I suggest to turn WHS auto-logon on. I suspect there is file corruption related to WHS having a console session open as well as WHSAdminConsole terminal services session at the same time.
I narrowed down the issue to the Cache folder. The folder become corrupts in a way that cause SBS to misbehave. With the Cache folder getting corrupt SBS http server will listen and accept connections but will keep them in open state. This will cause the WHS Console to get stuck and stop responding when opening the settings window, making it impossible to stop SBS without RDP to WHS and stopping it manually. At this point Squeeze mysql need to be stopped the the Cache folder get deleted or renamed. I speculate the reason this corruption happens only after a restart is that some file operations done in cache folder are delayed if there is another session until that other session is closed. This can also be related to the use of Application Data folder under Documents and Settings rather than a non user location (like Program Files) which makes more sense when running as a service. To reproduce make sure to use SBS (scan ans streaming) while there is a regular local console session, as well a remote WHSconsole session). Then restart.
do you still have the server.log file from when this happened? Might be helpful.
There is nothing in the server log file. There is one line each time the server is started. It does not seems the server recognizes there is an error condition.
By the way, is there a way to recreate the Cache folder, without reinstalling if I want to manually delete it?
*** This bug has been confirmed by popular vote. ***
Just to update this old issue. I learned that the the server will eventually operate normally, but it can take 30 minutes to few hours till it does. Perhaps it is mysql recovering an innodb database after an unclean exit? If so MySQL 5.5 has improved the recovery speed significantly. See here: http://blogs.innodb.com/wp/2010/04/innodb-performance-recovery/ In the tested scenario the improvement is from 7h38m to 13m56s of recovery time.