Alfresco becomes inaccessible

[konsultex]

I'm using Alfresco 2.9b on Windows 2003 server sp2 with Sun JDK 1.6.0_14 64 bit on a well sized server (Xeon quad core, 24 gb ram). I installed this in Windows from the Alfresco Tomcat bundle quite a long time ago, around the time this version became available (in 2008 I believe). In the last months I upgraded the JDK to the version I'm using now. I have a web application in the same Tomcat with Alfresco (a separate url) that works together with Alfresco. Alfresco talks to MySql and the application to MySql and Sql Server 2005 using JTDS. The users are authenticated by Alfresco which itself talks to Active Directory. Alfresco is used as the storage for the application and provides services in the form of web scripts for transitioning work flows, getting users, transfering files (via ftp), etc. but nothing really "heavy duty". There are currently about 40,000 documents in the storage which is about 25 gb in size. About 3 weeks ago I moved the storage (alf_data) to another directory on the same server due to lack of space in the original drive. No problems were detected after we started up again. The application has about 100 users but probably not more that 10 at any one time. The JDK options are generously sized for heap and PermSpace and we have not seem any OutOfmemory errors. Performance is acceptable for the application.

Out of the blue I now get incidents where the users can not login to the application, which happens because the application can not authenticate the user with Alfresco. I found that Alfresco itself is not available at that time, by which I mean that the /Alfresco url keeps you waiting forever with a blank screen. Without restarting Tomcat, if I use the Tomcat manager application I can restart Alfresco and the application and everything works again.

This started to happen about 2 weeks ago but as time passes it's getting more frequent and now happens several times a day. There might be a tendency to blame the problem on the new location of the storage, being the most recent change but I think this is not related although it happened at about the same time. I just mention this to supply all the facts.

Since I don't see anything strange in the logs I'm at a loss about where to look for a solution. My log may be hiding things though because I notice that after Alfresco starts, the alfresco.log file is cleaned so I really lose all the stuff during startup. Regarding the alfresco.log file, what part of Alfresco should I put into debug to try to get a better view?

Would my problem be an Alfresco problem (after all it does not even get to the login screen), a Tomcat problem, a JDK problem? Has anybody experienced this?

Thanks.

[konsultex]

In the meantime I tracked this down to a problem that seems to be related to the Alfresco ftp server. The application makes several ftp requests to Alfresco but after a certain amount (it does not seem to be constant) the ftp server seems to hang. Here is a log fragment (ftp in debug) showing this effect:

23:53:48,760 DEBUG [org.alfresco.ftp.protocol] FTP session request received from 172.16.0.6

23:53:48,760 INFO  [org.alfresco.filesys.smb.server.repo.ContentDiskDriver] Added file state for pseudo files folder (getinfo) - \

23:53:48,775 DEBUG [org.alfresco.ftp.protocol] Cwd ftp=/Alfresco/apd, share=Alfresco, path=\apd

23:53:48,775 INFO  [org.alfresco.filesys.smb.server.repo.ContentDiskDriver] Added pseudo files for folder (exists) - \apd\CONSTR\02005\54498308000138

23:53:48,775 DEBUG [org.alfresco.ftp.protocol] Cwd ftp=/Alfresco/apd/CONSTR/02005/54498308000138/6316, share=Alfresco, path=\apd\CONSTR\02005\54498308000138\6316

23:53:48,775 DEBUG [org.alfresco.ftp.protocol] Port open addr=/172.16.0.6, port=1681

23:53:48,775 INFO  [org.alfresco.filesys.smb.server.repo.ContentDiskDriver] Added pseudo files for folder (exists) - \apd\CONSTR\02005\54498308000138\6316

23:53:48,775 DEBUG [org.alfresco.ftp.protocol] Returning ftp=/Alfresco/apd/CONSTR/02005/54498308000138/6316/CONSTR~02005~54498308000138~6316~NF.pdf, share=Alfresco, path=\apd\CONSTR\02005\54498308000138\6316\CONSTR~02005~54498308000138~6316~NF.pdf

23:53:48,775 INFO  [org.alfresco.filesys.smb.server.repo.ContentDiskDriver] Added pseudo files for folder (exists) - \apd\CONSTR\02005\54498308000138\6316

23:53:49,760 DEBUG [org.alfresco.ftp.protocol] FTP session request received from 172.16.0.6


—> here it hangs until Tomcat is stopped

23:56:03,214 INFO  [org.apache.catalina.core.ContainerBase.[Catalina].[localhost].[/alfresco]] Closing Spring root WebApplicationContext

23:56:03,229 INFO  [org.alfresco.ftp.server] Shutting server FTP …

23:56:03,229 DEBUG [org.alfresco.ftp.protocol] FTP Close session, id = 93

23:56:03,261 DEBUG [org.alfresco.ftp.protocol] FTP Close session, id = 91

23:56:03,261 DEBUG [org.alfresco.ftp.protocol]  Error during transfer

java.net.SocketException: Socket closed

   at java.net.SocketOutputStream.socketWrite(SocketOutputStream.java:99)

   at java.net.SocketOutputStream.write(SocketOutputStream.java:136)

   at org.alfresco.filesys.ftp.FTPSrvSession.procReturnFile(FTPSrvSession.java:2145)

   at org.alfresco.filesys.ftp.FTPSrvSession.run(FTPSrvSession.java:4295)

   at java.lang.Thread.run(Thread.java:619)
‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍

So it looks like something happens in Alfresco right after this line:
23:53:48,760 DEBUG [org.alfresco.ftp.protocol] FTP session request received from 172.16.0.6

and before this line:
23:53:48,760 INFO  [org.alfresco.filesys.smb.server.repo.ContentDiskDriver] Added file state for pseudo files folder (getinfo) - \

I need to add that I tested Alfresco (same version) and the application on a different server and there it works fine.

Can somebody point me to where I can find a solution to this?

Thanks.