Notes frequently runs very, very slowly, even when opening a document in a local database. I have been able to correlate this performance degradation to ntaskldr. This task seems to kick in every time I replicate (every 3 minutes) and runs for a minute or two (I haven’t timed it). While it runs, it consumes 50%-70% of my CPU.
This problem started happening about 2 weeks ago. I had been running Notes 8.5 for some time before this problem began. The only change I can identify that correlates with the start of this problem is that I used the 8.5 designer for the first time that week.
In hopes of alleviating the problem, I deleted all the full text indexes I had on all my local databases, but to no avail. I reduced the number of databases I replicate frequently to those essential to my work, but again the problem persists.
I run the light Notes client (nlnotes), since the regular client causes my machine to crash frequently (as does Symphony, but that’s another story).
Any ideas how to fix this annoying problem?
Subject: ntldr CPU on replication
The same problem has suddenly appeared for my system as well. I have Notes 8.02 on XP – During replication of each database, it sits for some 40-50 seconds at “Searching for Database on ” – with ntsskldr running at 60%-90% with disk activity. I erased and re-replicated one database as as test – no change. I uninstalled and wiped the Notes directory, then re-installed Notes – no change.
Subject: I’ve never known this not to happen…
-
I’ve seen ntaskldr gobble up CPU since I started doing Domino over 12 years ago. For me the problem went away when I got my first hyperthreaded CPU … then ntaskldr could take up one thread and the machine stayed responsive using the other one. Slower, but not nearly so dramatic. I do know that R8.5.1 on a single CPU VM does not “freeze up” like this, though I don’t know if that’s because the plumbing has changed or if it’s because the VM isn’t really using a core, it’s using a virtual core, and that’s not the same thing.
-
Didn’t know this was replication. I always thought it was indexing. I’ve never heard of replicating every three minutes. That’s pretty darned short!
-
Main thing is, imho, this isn’t user correctable. Replication/indexing have to happen, but Lotus could make this process a little more UI-friendly than saturating the only core for it. (shrug)
-
You might see if you can decrease the priority of the ntaskldr process, but that’s a shot in the dark at best.
Hope this helps. Don’t know how it can, but I still hope it does…