[CalendarServer-users] Authentication Issues or Something Else?
Sean McAvoy
smcavoy at ripe.net
Thu Feb 19 02:00:02 PST 2009
I suggest checking /Library/CalendarServer/Documents/principals/
__uids__ and /Library/CalendarServer/Documents/calendars/__uids__ for
GUIDs that are no longer used.
And then you might want to run the proxy clean tool to clear up
entries in the proxy database that no longer have calendars:
http://trac.calendarserver.org/wiki/ProxyCleanup
On Feb 19, 2009, at 00:09, Don Kruse wrote:
> We have been using iCal/CalDAV server along with iCal clients now
> for about 2-weeks and we have had some of the same issues others on
> this list have reported but we have either been able to resolve
> these or at least understand them. Since Monday, 2/16 an issue has
> started to appear that is new and may or may not be directly iCal/
> CalDAV related. The problem appears to be with authentication.
>
> On the iCal/CalDAV server the DirectoryServices process is eating up
> CPU performance and response times for Calendar Server requests have
> started to become so long they are timing out for most but not all
> users. This was not an issue before as the DirectoryServices process
> typically was something under 10% of the CPU load at any given
> moment. We use an Open Directory (OD) server with Kerberos and this
> has been working well for the past 6-months. The iCal/CalDAV server
> is set up as "Connected to a Directory System" to get authentication
> from the OD. Our internal network appears to be OK, it is not
> saturated and is performing well. The OD server appears normal too.
>
> There is an error repeating in the CalDAV log about a single UID not
> being found, this error repeats so many times I wonder if requests
> for this calendar are causing the slow down. This error repeats
> several thousand times per minute or 59 times per second so I can
> see why this might be it. So how do I make this go away because this
> may be an issue on the client side right? Users are making requests
> for a calendar/user that no longer is available in OD. Also would
> there be any advantage to making the iCal/CalDAV server an OD replica?
>
> Where I'm at is that I don't know if this is truly and issue with
> our Calendar Server, the network, or possibly the OD server. Any
> insight is greatly appreciated.
>
> Here is some of the data:
>
> Calendar Server | Mac OS X 10.5.6
>
> This mostly one entry repeats in the caldav log over and over. I
> assume it is related to a user deleted from the OD but I'm not sure
> why this won't go away and it makes me suspicious this is at least
> partly what is delaying authentication requests. Below the log is
> the server performance snapshot showing the DirectoryServices PID at
> the top.
>
> caldav log:
> 2009-02-18 14:28:51-0800 [-] [caldav-8016] [OpenDirectoryService]
> Unable to find any record with GUID 77C3A32E-7315-4FD9-
> ABE4-1F3BDA78EAD0
> 2009-02-18 14:28:51-0800 [-] [caldav-8016] [-] u'No principal found
> for UID: 77C3A32E-7315-4FD9-ABE4-1F3BDA78EAD0'
>
> Processes: 72 total, 5 running, 67 sleeping... 271
> threads 14:09:14
> Load Avg: 3.37, 4.45, 4.75 CPU usage: 18.99% user, 8.77% sys,
> 72.24% idle
> SharedLibs: num = 7, resident = 45M code, 1388K data, 4448K
> linkedit.
> MemRegions: num = 8365, resident = 889M + 15M private, 1737M
> shared.
> PhysMem: 789M wired, 973M active, 36M inactive, 1806M used,
> 8434M free.
> VM: 3251M + 371M 20873(0) pageins, 0(0) pageouts
>
> PID COMMAND %CPU TIME #TH #PRTS #MREGS RPRVT RSHRD
> RSIZE VSIZE
> 26 DirectoryS 152.4% 6:48:49 13 115 164 9644K+ 584K 12M
> + 32M+
> 202 Python 9.5% 3:29:02 5 44- 453 99M- 1696K 101M
> + 116M-
> 204 Python 7.7% 3:43:18 4 39 402 80M 1696K
> 82M 97M
> 210 Python 7.0% 7:09:21 5 46 643 148M 1696K
> 150M 164M
> 200 Python 6.5% 3:13:29 4 39 364 69M 1696K
> 71M 86M
> 2040 UserNotifi 5.8% 42:09.88 2 87 93 1380K 14M
> 5136K 148M
> 203 Python 5.8% 2:36:24 5 46 376 73M 1696K
> 75M 90M
> 201 Python 5.8% 1:45:34 4 39 363 70M 1696K
> 72M 86M
> 0 kernel_tas 4.5% 24:12.76 65 2 405 7376K 0
> 329M 156M
> 465 top 3.9% 47:49.61 1 21 34 1480K 200K
> 2072K 19M
> 69 CoreRAIDSe 2.5% 6:05.67 2 67 35 408K 1500K
> 2196K 30M
> 43 Python 2.2% 8:30.92 1 15 219 11M 1672K
> 13M 28M
>
> Open Directory Server | Mac OS X 10.5.6
>
> This all seems really typical so I could not find any red-flags here
> on in any logs I scanned.
>
> Processes: 87 total, 2 running, 85 sleeping... 475
> threads 14:11:52
> Load Avg: 1.72, 1.35, 0.93 CPU usage: 12.65% user, 25.71% sys,
> 61.63% idle
> SharedLibs: num = 2, resident = 40M code, 0 data, 3608K
> linkedit.
> MemRegions: num = 11086, resident = 551M + 16M private, 54M
> shared.
> PhysMem: 253M wired, 700M active, 129M inactive, 1083M used,
> 965M free.
> VM: 3486M + 129M 130707(0) pageins, 20334(0) pageouts
>
> PID COMMAND %CPU TIME #TH #PRTS #MREGS RPRVT RSHRD
> RSIZE VSIZE
> 43853 slapd 18.5% 89:50.06 9 51 124 13M 624K
> 16M 200M
> 3761 AppleFileS 13.3% 3:25:40 202 492 1454 118M 3060K
> 120M 324M
> 46661 top 12.7% 0:02.60 1 21 33 1244K 200K
> 1836K 19M
> 0 kernel_tas 10.4% 6:12:47 51 2 833 14M 0
> 246M 157M
> 25 DirectoryS 2.9% 2:46:42 8 91 113 4744K 636K
> 7196K 25M
> 8805 SecurityAg 1.8% 3:11:20 3 93 100 1700K 4708K
> 6244K 120M
> 419 smbd 0.3% 2:20.13 1 17 53 128K 2392K
> 2136K 23M
> 82 named 0.3% 14:37.08 1 14 107 24M 200K
> 25M 38M
> 8797 WindowServ 0.1% 15:46.87 4 105 63 4996K 992K
> 6628K 81M
> 81 mysqld 0.1% 15:42.35 9 40 55 53M 200K
> 55M 119M
> 1 launchd 0.1% 3:29.61 3 140 26 240K 320K
> 600K 19M
> 109 emond 0.1% 8:44.22 1 32 23 364K 464K
> 1824K 27M
>
> _______________________________________________
> calendarserver-users mailing list
> calendarserver-users at lists.macosforge.org
> http://lists.macosforge.org/mailman/listinfo.cgi/calendarserver-users
More information about the calendarserver-users
mailing list