NT - Slow Service Start (NSRD)

[ShawnL]

Has anyone run into an issue where it takes forever for your Backup and Recover service to start? How long does most of yours take? From the beginning (we started with 5.5.2) it took about 15 minutes to start (which still is long). Now after many other problems and after upgrading to 6.0.1, it now take 30 minutes to start which is unacceptable. It takes us hours and hours, just to troubleshoot a problem if we have an issue. We have been told it was our NSR.RES file that was corrupt and it is what's doing it, but support has never been able to back that up. They have looked at our file and can never find anything wrong with it, and I have previously recreated it and still got the same results. I have also been getting a condition recently (last 2 or 3 months) where either the NSRD Dr. Watson's or I have to stop it for some other reason, and when I try to start it, it won't come up. Sometimes, I can just do an MMRECOV and use another RES and that works (although support can not find any thing wrong w/ the RES). And other times, I can try 3 different RES files, and they don't work, but go back to the original one that wasn't working, and it works. Plus, if it happens one day and I get a RES file to work, if it happens again another day, and I try to use that same one that worked again, it doesn't work. There just isn't any rhyme or reason. Now they are trying to tell me my client file indexes are corrupt. I'm not really buying that. Anyone have any similar experiences or can shed any light or give ideas, please let me know.

 

[ukraine12]

when starting up legato, the product executes some check, which normally you should be able to follow in the daemon.log on your machine.
the first check is in the media database.
after that he starts checking all the clients one by one.
if you have very large clients(check the index size) it can take long to end. if there are any client corrupt you will see it in the daemon.log when he is checking the specific clients(to repare this execute : nsrck -F client-name).
Last time we restarted legato 5.1 it took us 13 minutes for
123 servers.
Considering the size of our clients i don't think this is too bad.

 

[x2shy]

what hardware are you runing and how fast is the server? We have about 30 clients per Legato server and it only take about 3-4 minutes max to stop and start services. I think you might want to check your server harware possibly a full rebuild with the latest SP?

 

[ShawnL]

Well, if anyone is interested, here is the latest. It doesn't have anything to do w/ our hardware or number of clients or etc. We have 95 clients, NT4 SP6a, dual P3 733, 640 megs ram. What they had me do was kill my RES file and totally recreate it from scratch. when I started doing that, the start time for the service was all the way down to 20 seconds. I was totally done with it except for one part, I had done the groups, pools, directive, schedules and etc. The only thing I didn't do was add the storage node affinity to the clients (the section of the client definition where you tell it which storage node to backup to). As soon as I started adding some of those, the time climbed, to 1 minute, to 4 and on up. I talked to support and they told me to just add one definition to each client (the storage node I want it to go to and the server), previously I listed all storage nodes. I did it this way and I'm down to 9 minutes for start up, which is a million times better than 30, but it still doesn't make sense why that setting would matter. They are still looking into it. They are also looking into a problem I'm having w/ more MMD's running on the storage node than should be. Then no tapes load and stuff like that. That also might be related to this, but we'll see.