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

RE: Subversion client hangs during long operations

From: Jason Erickson <jasone_at_dbo2.com>
Date: 2007-06-22 00:56:44 CEST

Some additional information: Since I can do web-browsing just fine, I took a
chance that the problem was Windows Vista and the svn protocol. I got
Subversion up and running on Apache so I could use http, but the behavior is
the same.

I'm getting really stuck here. Not using Subversion is not an option. Not
using Vista is an unattractive option, but not in the long run. Should I
log this as a bug?

-----Original Message-----
From: Jason Erickson [mailto:jasone@dbo2.com]
Sent: Wednesday, June 20, 2007 9:04 PM
To: rsivak@istandfor.com
Cc: users@subversion.tigris.org
Subject: RE: Subversion client hangs during long operations

I have disabled IPv6, antivirus, and Windows Defender. None of that helped.
I don't have httplook installed. According to the Windows help, you can't
turn off indexing anymore.

I ran WireShark and captured until my process hung. Sometimes it hangs
quickly and sometimes it takes several minutes. This time it hung quickly.
I don't really fully understand what I'm looking at because I'm not fluent
in TCP.

Things start off fine with lots of [SYN], [SYN, ACK], [ACK], [PSH, ACK] and
[FIN, ACK] messages. The [PSH, ACK] are the most common.

We pretty quickly (.75 seconds) get a line that says [TCP Window Update]
51108 > 3690 [ACK] Seq=273 Ack=30474 Win=17408 Len=0

Things look OK for about a quarter second except we have lots more [ACK]
than [PSH, ACK] if that means anything. Then we get another TCP Window
Update, again with Seq=273.

This goes on for a while - several messages look fine and then a TCP Window
Update - for a couple of seconds. Then we get a line that says [TCP
Zerowindow] 51108 > 3690 [ACK] Seq=273 Ack=225271 Win=0 Len=0. After that
one we get two more TCP Window Updates in a row.

We go back to the back-and-forth pattern for another half second, get
another TCP Zerowindow, again with Seq=273. Immediately after that we get
two more TCP Window Updates in a row. Then we get a line that says [TCP
Retransmission] 3690 > 51108 [ACK] Seq=267407 Ack=273 Win=17248 Len=1460.
Then we have [TCP Dup ACK 432#1] 51108 > 3690 [ACK] Seq=273 Ack=270327
Win=17248 Len=0 SLE=267407 SRE=268867

From now on we just alternate between TCP Retransmission and TCP Dup ACK
432#n.

Does this give anyone any clues? Would it be easier to attach the capture
file from WireShark? (I'm not sure of the mail-list etiquette on that.)

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Fri Jun 22 00:58:05 2007

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