[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: multiple access bog down on HP-UX 11i

From: Ryan Hunt <rhunt_at_hp.com>
Date: 2003-11-14 18:43:01 CET

So one thing I have noticed, and am not sure how related to this
problem is that the svnserve -t process does not reliably die on the
server after the client is done with the transaction. Whether the
client exits gracefully or not the server side process continues.

Again this is using svn+ssh protocol.

Does anyone know of any issues that might be causing this?

-Ryan

On Thursday, November 13, 2003, at 02:35 PM, Ryan Hunt wrote:

> So I spoke too soon when I said it does not occur with a
> checkin/checkout combo.
>
> I just got a report of the same thing happening. In the following
> listing you can see how the rate that the log files are created slows
> right down... There are quite a few logfiles created on 13:55, and
> fewer on 13:56, and from there it is all down hill.
>
> -rw-rw-r-- 1 dprosser users 1040536 Nov 13 13:55
> db/log.0000159496
> -rw-rw-r-- 1 dprosser users 1042794 Nov 13 13:55
> db/log.0000159497
> -rw-rw-r-- 1 abowman users 1042794 Nov 13 13:55
> db/log.0000159498
> -rw-rw-r-- 1 dprosser users 1040536 Nov 13 13:55
> db/log.0000159499
> -rw-rw-r-- 1 dprosser users 1042794 Nov 13 13:55
> db/log.0000159500
> -rw-rw-r-- 1 dprosser users 1042794 Nov 13 13:55
> db/log.0000159501
> -rw-rw-r-- 1 dprosser users 1040536 Nov 13 13:55
> db/log.0000159502
> -rw-rw-r-- 1 abowman users 1042794 Nov 13 13:55
> db/log.0000159503
> -rw-rw-r-- 1 abowman users 1042794 Nov 13 13:55
> db/log.0000159504
> -rw-rw-r-- 1 dprosser users 1040536 Nov 13 13:55
> db/log.0000159505
> -rw-rw-r-- 1 dprosser users 1042794 Nov 13 13:55
> db/log.0000159506
> -rw-rw-r-- 1 dprosser users 1042794 Nov 13 13:55
> db/log.0000159507
> -rw-rw-r-- 1 dprosser users 1040536 Nov 13 13:55
> db/log.0000159508
> -rw-rw-r-- 1 abowman users 1042794 Nov 13 13:55
> db/log.0000159509
> -rw-rw-r-- 1 dprosser users 1042794 Nov 13 13:55
> db/log.0000159510
> -rw-rw-r-- 1 dprosser users 1040536 Nov 13 13:55
> db/log.0000159511
> -rw-rw-r-- 1 dprosser users 1042794 Nov 13 13:55
> db/log.0000159512
> -rw-rw-r-- 1 dprosser users 1042794 Nov 13 13:55
> db/log.0000159513
> -rw-rw-r-- 1 dprosser users 1040536 Nov 13 13:55
> db/log.0000159514
> -rw-rw-r-- 1 dprosser users 1042794 Nov 13 13:55
> db/log.0000159515
> -rw-rw-r-- 1 dprosser users 1042794 Nov 13 13:55
> db/log.0000159516
> -rw-rw-r-- 1 dprosser users 1040536 Nov 13 13:55
> db/log.0000159517
> -rw-rw-r-- 1 dprosser users 1042794 Nov 13 13:56
> db/log.0000159518
> -rw-rw-r-- 1 dprosser users 1042794 Nov 13 13:56
> db/log.0000159519
> -rw-rw-r-- 1 abowman users 1040536 Nov 13 13:56
> db/log.0000159520
> -rw-rw-r-- 1 dprosser users 1042794 Nov 13 13:56
> db/log.0000159521
> -rw-rw-r-- 1 dprosser users 1042794 Nov 13 13:56
> db/log.0000159522
> -rw-rw-r-- 1 dprosser users 1040536 Nov 13 13:56
> db/log.0000159523
> -rw-rw-r-- 1 dprosser users 1042794 Nov 13 13:56
> db/log.0000159524
> -rw-rw-r-- 1 dprosser users 1042794 Nov 13 13:56
> db/log.0000159525
> -rw-rw-r-- 1 dprosser users 1048515 Nov 13 13:59
> db/log.0000159526
> -rw-rw-r-- 1 dprosser users 1048562 Nov 13 14:04
> db/log.0000159527
> -rw-rw-r-- 1 dprosser users 1048262 Nov 13 14:13
> db/log.0000159528
> -rw-rw-r-- 1 dprosser users 1048523 Nov 13 14:18
> db/log.0000159529
> -rw-rw-r-- 1 dprosser users 941327 Nov 13 14:24
> db/log.0000159530
>
> notice how long it takes to write that last 5 log files...
>
> -Ryan
>
>
> On Thursday, November 13, 2003, at 01:50 PM, Ryan Hunt wrote:
>
>>
>> On Thursday, November 13, 2003, at 01:32 PM, Philip Martin wrote:
>>
>>> Ryan Hunt <rhunt@hp.com> writes:
>>>
>>>>>> The import is killed mid stream.
>>>
>>> What does this mean? Do you manually abort the import? Does the
>>> system kill the client or the server? Does the server abort?
>>>
>> I manually abort the process because at the slow data rate it would
>> take nearly 3 weeks to complete the transaction.
>>
>>> If the process accessing the repository (the server) is killed or
>>> aborts while it is holding BDB locks then it is quite likely that the
>>> repository will have problems.
>>>
>> Exactly, so after the kill I run a svnadmin recover.
>>
>> I can then resume checkouts or imports, but not both simultaneously.
>>
>> This does not appear to be a problem with checkins and checkouts,
>> only imports and checkouts.
>>
>> --Ryan
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
>> For additional commands, e-mail: users-help@subversion.tigris.org
>>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: dev-help@subversion.tigris.org
>

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Nov 14 18:39:50 2003

This is an archived mail posted to the Subversion Dev mailing list.