[SOLVED] slow performance in one account

[gnyce]

Just finishing up our migration, and I have someone who says the calendar, for example, is quite slow... to create, but especially to delete. I verified that this is correct... both IE7 and FF2/3. It can take 5-10 seconds to delete a simple test-appointment. Thinking it was a PC issue, I looked... intel 2.4 ghz, 1 gig RAM, not horribly stressed. Network link is ok. Hmmmm.... so I grabbed a laptop - a decent core2 duo. Using the admin UI, I went into his account... verified, so should not be a PC issue. So tried another account - his manager has a ton of appts. Much faster. Weird, so I tried a third account (all from the same network link and laptop), that was also much speedier. Went back to the original account, still slow on the calendar.

So what could be in one particular account to make it slower? He has about 700 meg of mail, versus much larger on the two other accounts I tested (both over 1 gig). Thought about re-indexing the mailbox, but no idea if that would help. Which logs should I be looking for in such a case? Some settings that he made? Even changed the theme back to the default. Not sure where to look first, as I have not seen something like this so far....

[gnyce]

Turns out - from the migration, I guess - there are three folders in his calendar (nested two-deep), with another Calendar listed under there. No way to delete the folders from the web client, but I did delete the 2nd calendar. Probably a cmd-line way to delete these folders in there... will check the wiki for cmd-line tools... maybe _this_ has something to do with the slow perf. times?

[mmorse]

Hm, I have sub-cal's & don't notice a slow down, might be the stuff in the calendars themselves, but you could use the zmmailbox command if that's what you're thinking it is...

Usage:
su - zimbra
zmmailbox -h
zmmailbox help folder
zmmailbox -z -m user@domain.com getAllFolders

Warning: Export an ics of the calendar first before doing the following which removes data - REST url is like: server/zimbra/home/uesr/calendar/subcal.ics or you can visit the account > preferences > calendar tab.

zmmailbox -z -m user@domain.com emptyFolder /Calendar/subcal
or
zmmailbox -z -m user@domain.com deleteFolder/Calendar/subcal


UI deletion will be handled in Bug 27211 - Support user creation and moving of nested calendars (currently you can create sub-cal's using other clients)

[gersh]

Just upgraded to 5.0.7 in a multi-store environment. Calendar rendering is beyond slow, with "server timeouts" appearing before the calendar appears. Users cannot create new appointments without timeout message appearing.

Newly created calendars are OK. It's the default "Calendar" that will not render. We noticed this in several users who's calendar contains many events/appointments.

Anyone notice this?

gersh

[gnyce]

Can you export and delete the main calendar, and then re-create it? I'm thinking about exporting into an .ics, deleting his default Calendar, re-creating it, and then importing. Would that help? I have no idea, but not sure what other "next steps" there should be.

gersh - sorry, we're a single server, and my problem seems to be with just one account, but as we've just migrated (finishing up now), there could be others and maybe they think this is normal??

[mmorse]

See the REST url above for example of exporting, for the /Calendar you need to use the emptyFolder command as it's immutable (you can't use deleteFolder).

[gnyce]

I ended up emptying his Calendar (thanks mmorse for the correct syntax on that). The user had about 270 items - again, not a large amount - but clearing out the calendar made a huge performance difference in terms of speed of creating and deleting. We'll see if it slows down again as he inputs appointments.

I thought about importing the .ics file that I had from exporting his calendar, but he chose to just re-enter items.

I checked the number of appts he had (270+) versus some other known-large calendars... one had 1,200+ items, and yet was very much faster/response. I have the .ics file if anyone wants to take a look, but I have no idea how you would check for corruption or problematic entries. This was migrated from GroupWise using the GWMigration utility - but again, there have been no other reports of something similiar (so far).

[kbigger]

Have user with 1600+ appts and meeting that's developed a super slow calendar. One question, I seem to recall from previous experience that dumping and reloading cal meeting requests resends all the invites. Is that correct and how does one stop that from happening.