Monday, June 21, 2010

Bit by the obvious

I hate it when I spend hours troubleshooting something that's right under my nose and should be "obvious" to me.  Two hours wasted tracking down an error that I should have hit on right away.  Instead I discovered that there's only 2 references to the error when googled and one of them wasn't obviously resolved either.  So I'll document it here in hopes that someone else may find it in a Google search and I'll have saved them some time.

The odd part is that this popped up in my newly rebuilt and clean testbed.  All virtual servers, some on OES-Linux, some on SLES 10, and some on NetWare 6.5.  (I need a complex test bed to mirror what we're doing in production).  Opening up ConsoleOne to get at the GWIA object - all so I could add on an HTTP user name and password - gave me the following error:

Unable to locate the file "\/wpgate/gwia/gwac.dc".
Full administration for the gateway will not be available until installation is complete.

A search of the Novell Knowledgebase turned up nothing on this error, there's one reference to it in the forums, which I'll have to go post a response to as well.

The cause of this error?  The domain object (not the MTA) had the UNC path set to a "\".  Where it picked this up from I have no idea as I created the domain during a fresh, new install of GW 8.01 HP1.  But there it was, as shown below, all mucked up.  Switch that to /mail/grpwise (my domain path in my testbed) and life is all groovy again.


Sorry for the sloppy graphic, but I don't have screen capture software loaded on the virtual boxes.  Usually I have these odd issues in the production systems, not the test.  Odd, but a fun one to deal with on a Monday.

Monday, June 14, 2010

New GroupWise Hiccups

For awhile now, I've been warning the powers that be that our eDirectory still has issues that can only be fixed by creating a new tree.  We had the perfect opportunity when we migrated to Linux servers and all new drive mappings, but I was nixxed on the decision.  The chant of "but we've had this tree since we went to 4.0" (which was badly botched BTW) was consistent and no one would listen to me.  Not a big surprise in a place where the motto seems to be "but we've always done it this way".

Because of this, I'm never surprised when there's an issue with a newly created account not getting it's GroupWise user database created properly.  It took me awhile to figure out why I was getting C05D errors sent to my mailbox with message attachments that I couldn't read.  I had set my POA logging to verbose and was seeing C05D messages related to another user ID.  I dug around in the logs for the C05D information and finally hit on the fact that it was related to an error I was seeing in my mailbox statistics log and finally put the two together.  Which was really good as the long time user of the other account that was being "blamed" would have had a fit if I tried a structural rebuild or re-create database on his account.

I found errors in my weekly GWCheck process logs - namely the one that captures mailbox statistics.  The error I found in the statistics log looks like this:  (with the letters fid indicating the particular users FID number)

          Checking user = aaa1 (fid) 1214/1448 - 0 bytes
                   Database not found - /mail/postofc/ofuser/userfid.db


I usually "fix" this issue by moving the user to a different PO and moving it back.  Not this time.  Oh it worked great moving it to my "all fubar accounts" post office, but moving it back failed miserably.  In fact the move couldn't finish because the POA couldn't create the new database for whatever reason.  Nope, no errors in the log - which was still set to verbose, we're not out of space, and the domain and agents have full rights to the drive space.

In an vain attempt to resolve the problem, I copied the user database over to the correct server.  This allowed the move to complete, but gave an error anytime a GWCheck process was run.  Not unexpected since I did copy it over from the second PO to the preferred PO.  The error I would get is an error 44 with the following message:

           Error 44 - Database userfid.db is invalid due to security breach!
             - Verification HOST_ID is "gwpo2", should be "gwpo1"

Novell's fix for this it to restore the database from backup, but since there never was one on "PO1" to begin with there was nothing to restore.

I ended up "fixing" it by first checking that the account was inactive (meaning it's never been logged in), then I deleted just the GroupWise account and readded the user to the correct PO in the GroupWise tab in ConsoleOne.  I didn't need to adjust the FID as there was nothing to recover, so I left the account with a new FID and ran a GWCheck just to verify I no longer received any errors.  All was fine and dandy.

Now I just have to wait and see what other bizarre things will occur with this account. DSRepairs show no errors on the specific object and our eDirectory Health Check shows our tree to be healthy - for the most part.  There are things I cannot fix without creating a new tree, so I'll just keep patching with bandaids until it dies in some horrible, complicated fashion.

Wednesday, June 2, 2010

Novell Pulse, BrainShare Amsterdam, and TTP Conference USA

I've discovered that my first 5 invites for Pulse didn't get to all of my invitees.  Surprise, surprise!  So, of course I complained (everyone knows I'm mouthy).  Today I logged in and found 10 new invites available.  I don't know why and I'm not asking, I'm just taking advantage of them.  And I've found that as long as I invite one person at a time, instead of a whole list separated by commas, it seems to work.  Now, just to figure out what to use Pulse for.  Any ideas?

This whole Novell Pulse / Google Wave stuff is a bit beyond me at the moment.  Not that I can't figure out the "how", it's more of the "why" to use it.  I'm still working on that bit, but am honored Novell's including me in this instance of Pulse.  I'm thinking it has broader uses for where I work than for the more social aspect of it, it's a new technology, so I'm assuming the training wheels are still on it.  It will be interesting to watch this evolve.

BrainShare Amsterdam has wrapped up and appears by all accounts to have been a resounding success.  Nice to have it back in Europe again, although I'll miss seeing friends at the Utah version.

TTP's annual conference is back in Provo from July 11 through the 16, 2010.  Peter is working on the schedule now, so if you've got some good technical presentation you'd like to give, let him know ASAP.  I'll be missing out on it this year (no $$ for travel).  I'm glad I had a chance to catch the TTP Summit prior to BrainShare this year.

On the tech front, not much is happening.  I finally figured out some of the eDirectory oddities I've been seeing aren't eDir after all and is all our LDAP server's fault.  So no techie points there to tell you about.

I'll be working on my NCE training this summer (all self-study) along wtih CLA/CLP and MCSE training.  SharePoint is here (unfortunately), so I've offered to try to take on the administration of that beastie.  We'll see if I'm actually allowed to do so, as it's all yet to be determined.