OWA down

[Ryan]

Another Exchange question while we're on the subject.

 

We run Exchange 2003 on our PDC. Occasionally webmail becomes inaccessible, normally when the server's been up for more than a week... Is there any way i can restart/enable it (by services or otherwise) without rebooting the server?

 

This may be a sympton of something deeper, dunno. Ideas please lads (oh, and ladies)

[JamesC]

Sounds like it could be your IIS WWW Publishing Service that has a problem. I would create a .bat file with the following lines:

 

net stop W3SVC

net start W3SVC

 

Save it somewhere on your c:\ and then create a scheduled task to run it every night at 4am.

 

EDIT:/ of course you can and should test this first by waiting for the OWA problem to occur and then opening a command prompt and typing net stop w3svc and then net start w3svc and see whether it is working again.

 

I would also suggest looking in the event log for errors, specifically relating to IIS etc because you should be able to resolve this without resorting to the .bat file above. The .bat should only be considered a temporary fix.

Edited February 21, 2008 by JamesC

[apeo]

We run Exchange 2003 on our PDC.

 

Not a good plan...

[Ryan]

Thanks for that dude. It's happening just now (web access dead) but the restarting of w3svc didn't resolve it i'm afraid

 

Not a good plan...

I know, i know. We don't have the space, money or need for separate servers per service though.

[apeo]

You getting any error messages? Can clients get to their mailbox using another client?

[Ryan]

Mostly all is well - they can get in using the full Outlook client program. Web access both internally and externally is dead.

 

Trawling through logs atm...

[apeo]

While you are checking that out, try restarting the exchange virtual server i.e. HTTP protocol in System Manager.

[srochford]

Thanks for that dude. It's happening just now (web access dead) but the restarting of w3svc didn't resolve it i'm afraid

 

 

I know, i know. We don't have the space, money or need for separate servers per service though.

 

 

Try doing iisreset - this will restart the iisadmin service as well which might help.

[Ryan]

Tried both those (iisreset and the HTTP virtual server), still no joy

[apeo]

So are the clients getting usual 'Page cannot be displayed' gumph or are you actually getting a useful error message?

[Ryan]

Firefox says: The connection was reset

 

IE (through ISA 2006) says: The page or file cannot be displayed

Reason:

The connection to the specified web server has been lost.

Technical error details:

 

Error Code 502 Proxy Error: The specified network name is no longer available.

Date: 21/02/2008 15:46:06 [GMT]

[Lee_K_81]

Firefox says: The connection was reset

 

IE (through ISA 2006) says: The page or file cannot be displayed

Reason:

The connection to the specified web server has been lost.

Technical error details:

 

Error Code 502 Proxy Error: The specified network name is no longer available.

Date: 21/02/2008 15:46:06 [GMT]

 

DNS issue? Have you checked for any conflicting ip addresses? Does the same server also handle DNS / DHCP?

[Ryan]

The server Exchange is on also does DNS and DHCP yep. Everything else seems to run fine (as far as i'm aware...) Just this periodical niggling webmail problem

[Lee_K_81]

Theres no other servers in the domain also handling DNS? Just wondering about replication delays. At one point i was having real issues with AD replication when my predecessor set up a second DC. Ballsed it up, reformatted and started again without demoting the server. Just deleted the server from AD. It was only when we got exchange and had problems with connections dropping, services stalling, etc. that we found hidden away in the schema a reference to the previous instance of the DC. Once this was deleted exchange has run like a dream.

[CyberNerd]

Check disk space and mailbox store, We had an exchange box that would start falling over if it approached the mailbox store limit. First OWA would break, then IMAP then the MAPI

[apeo]

I assume it still doesnt work even if you tried going directly to the server instead of going through ISA i.e. on the server itself try to get to owa but dont use a proxy to get to it.

[Ryan]

Morning lads,

 

Thanks for all the suggestions. In the end i rebooted the bugger this morning. Bosh, sorted. I'll keep an eye on it though, as it is a recurring (and annoying) problem. Evidence seems to point to BackupExec's Friday night backup, so this may get revisited.

[steve]

We have a similar problem I've never managed to track down.

 

Our OWA uses a dedicated NIC which every few hours becomes unavailable via our external ADSL line but still works internally

 

I have a scheduled task to restart the NIC:

 

netsh interface set interface "exchange" DISABLED
netsh interface set interface "exchange" ENABLED

[Ryan]

Posting a solution (of sorts) to this after finally diagnosing it.

 

I've been in WSUSadmin for a couple of days configuring and fiddling, and after coming back from lunch, the admin webpage (WSUS 2) was down. So i checked Outlook web mail access, and sure enough, that was dead too.

 

Long story short, it turns out that Backup Exec's rxWriterSvc service was chewing through NPP (non paged pool memory) resulting in IIS refusing any further connections. To check this, use Task Manager's Processes tab and do "View > Select Columns...". Add Non-Paged Pool and sort on that field to find the offender.

 

Problem was remedied by running "net stop rxWriterSvc" and "net start rxWriterSvc". Then i restarted IIS. Everything hunky-dory now, access restored to both WSUS and OWA.

 

Hope this helps someone.

[apeo]

Glad you found the cause of the problem... odd that veritas is doing that tho..