We recently moved our main Domino server to a freshly installed Solaris 9 MU2, 64 bit, dual processor machine. It seemed like a good time to upgrade to Domino 6.0.1, since we are also rolling out a server at a new location. The upgrade went exceptionally smooth, and until the system came under load, it appeared to be running fine. Except for the IT staff, the clients are still Notes 5.0.x. Mail templates are still the Notes5 ones.
After a day or so, we started noticing mail folders disappearing, and server messages like “Database is corrupt - Cannot allocate space”, “document UNID… is invalid and has been deleted”, “Document had invalid structure”, and one about dbfolder.c, but my log got overwritten on the last crash. As I read in another post, updall, fixup, compact, all seem to get things running again, but overall things are getting worse instead of better.
I’m trying to get registered to submit a support request, and I guess it’s a testament to Notes 5 that I have not needed to be registered with IBM until now. I’m considering moving it all back to Notes 5.0.x, but I’d rather not. Is anyone running 6 or 6.0.1 on Solaris? Good, bad, or ugly?
Any ideas? I know have not given enough detail for specific help. My plan is once I get a stable system back, to run fixup and compact on everything to make sure everything is R6 DB format, clean out the notes.ini of all the stuff that has built up over the years, and probably turn off the LEI transactions running on a 5.x machine which cause a high network load on the 6.0.1 server. Then I need to make sure that I get the server logs scrolled away so there’s something to look at.
Subject: Domino 6.0.1 on Solaris 9 not ready for primetime?
Just a followup. Over the last month the Notes 6 server on Solaris crashed 2 times with no real clues as to why. Starting a 4:00 PM yesterday, it hs started crashing on average once an hour. We have sent NSD files, etc., to IBM/Lotus, and they are looking at them, but no news yet.
Subject: RE: Domino 6.0.1 on Solaris 9 not ready for primetime?
Did you get a resolution to this? I certainly hope so. We’re contemplating an upgrade from 5.0.12 to 6.0.1, and this thread, along with the one from Debra Alioto, has me a little concerned.
Subject: RE: Domino 6.0.1 on Solaris 9 not ready for primetime?
I think we have had 3 major issues.
We had some database corruption that was apparently due to the 6.0.1 server not being 100% compatible with the 5.x ODS. Compacting all the databases to the 6.x ODS fixed this.
We had a series of hard crashes that were very frequent. IBM/Lotus support had me add a series of debug statements related to full text indexing. I sent them several crash dumps and then decided to nuke all the .ft directories and let the server rebuild them. Those crashes did not return, so I hope it was a compatibility issue between indicies generated with 5.x and the 6.0.1 server.
We continue to have infrequent server hangs. The database server stops responding to client access, but other tasks like SMTP continue to run. I have sent them 2 sets of dumps, and they believe it has something to do with semaphores, but I do not have a resolution yet. It has been escalated to the Domino on UNIX support group. I’ll try to remember to post back here when I do.
I am concerned enough that I have been delaying a rollout of 6.0.1 on Linux to a new location. It’s a location that is not so enthused about having their email moved to Notes, so problemrs like this could be really bad PR.
My advice if you do go forward, is to make sure that all your databases have been rebuilt to the 6.0 ODS and that you nuke and let the server rebuild all the ft indices. I still have the server hangs as an open issue.
Subject: RE: Domino 6.0.1 on Solaris 9 not ready for primetime?
We are experiencing the same types of server hang issues. Our log.nsf corrupts nearly once a week - closer to 1 every 8 or so days. Or the server will just stop responding with no NSD errors. Both servers are running 6.0.1 - one on NT4 (Ya, I know upgrade) the other on W2k server.
Subject: RE: Domino 6.0.1 on Solaris 9 not ready for primetime?
Lotus engineering claims our bug is only on Solaris. I have installed a hotfix from them and the early indication is that it has fixed the problem. It’s been about a week without a crash, but I won’t believe it until 3 weeks…
I’ll have to keep any eye out for your Windows server problem. I still have a W2K Domino server yet to upgrade. I suggest opening a call with Lotus support.
Subject: RE: Did you run Compact to upgrade the ODS on all dbs to 4.3?
I could be mistaken, but I thought the R6 server was supposed to be able to use the R5 (41:0) ODS. I have been compacting the mail databases when the users are idle, and it seems to be doing better now. The other strange thing I found was a DEBUG_BTREE_ERRORS line in the notes.ini. I don’t remember why it was in there. I removed it.
If the ODS was the problem, then I’m relieved! I just wish that with the old ODS it had either worked perfectly, or said “hey stupid, you need to compact before writing to this database.”
Subject: RE: Did you run Compact to upgrade the ODS on all dbs to 4.3?
mark,#1 - yes you can keep R5 formatted DBs on an R6 server. In fact, if one was updated accidentally, you can revert it back to an R5 structure from your domino 6 server…
In the admin client, right click on it, and compact it, but CHECK THE BOX to keep it as an R4* style - which acutally does a Compact -r on the database. This text MAY be changed in the Dom6 admin client, but that’s what it said when I ran it from a R5 Admin client ON the Dom6 server and it worked fine.
#2 - DEBUG_BTREE_ERRORS is a NEW - INFORMATIONAL statement … no need to worry. It’s been there (the variable) since like 5.05 but we didnt get the message, now they want us more in formed… go figure. but its no need to worry.
Subject: RE: Did you run Compact to upgrade the ODS on all dbs to 4.3?
It appears that I was geting corruption in the mail databases when that were R5 ODS, and I am not when then are R6 ODS. It’s supposed to work either way, so maybe it’s a bug, or maybe something else was the cause.
In any case, may system appears to be stable, so I’m alot happier.
Subject: They don’t specifically mention it but I remember it being required for the R4->R5 upgrade.
I remember getting all those kind of errors on 1 server R4->R5 where I did not run the compact task. If you need to keep a file in ODS 4.1, I would recommend that you have it named xxx.NS5
