issue with svn 1.2.0 client against 1.1.x server

NAVIGATION

CATEGORIES

Development

REFERRENCE

If you have some good idea, Cilck and mail to us please...

LINKS

Home » Development

issue with svn 1.2.0 client against 1.1.x server
8 answers - 1155 bytes -
Hello!
With subversion client version 1.2.0

>time /usr/local/svn12/bin/svn list https://openib.org/svn
branches/
gen1/
gen2/
tags/
trunk/
0.395u 0.067s 1:53.62 0.3% 0+0k 0+0io 0pf+0w
While with client 1.1.4 against the same server I have:

>time svn list https://openib.org/svn
branches/
gen1/
gen2/
tags/
trunk/
0.019u 0.007s 0:04.29 0.2% 0+0k 0+0io 11pf+0w
So subversion 1.2.0 is taking 2 minutes where 1.1.4 takes 4 seconds
to contact the same remote repository.
This observation was independently confirmed by another svn user.
Running strace on versions 1.2.0 and 1.1.4, I see that subversion 1.2.0
does around 17000 system calls, while 1.1.4 only around 750.
Further, most of the time while client 1.2.0 appears stuck,
it is actually performing read/write/select calls, presumably from the socket.
Both clients are built from source with ./configure
The openib.org server is AFAIK 1.1.x
This problem prevents me from changing my client to 1.2.0.
Is this a known issue?
Thanks,
No.1 | | 1315 bytes | |
Hmm I saw similar behavior the other day, but assumed it was some sort of
fluke.

"Michael S. Tsirkin" <mst (AT) mellanox (DOT) co.ilwrites:

Hello!
With subversion client version 1.2.0

>time /usr/local/svn12/bin/svn list https://openib.org/svn
branches/
gen1/
gen2/
tags/
trunk/
0.395u 0.067s 1:53.62 0.3% 0+0k 0+0io 0pf+0w

While with client 1.1.4 against the same server I have:

>time svn list https://openib.org/svn
branches/
gen1/
gen2/
tags/
trunk/
0.019u 0.007s 0:04.29 0.2% 0+0k 0+0io 11pf+0w

So subversion 1.2.0 is taking 2 minutes where 1.1.4 takes 4 seconds
to contact the same remote repository.

This observation was independently confirmed by another svn user.
Running strace on versions 1.2.0 and 1.1.4, I see that subversion 1.2.0
does around 17000 system calls, while 1.1.4 only around 750.
Further, most of the time while client 1.2.0 appears stuck,
it is actually performing read/write/select calls, presumably from the socket.

Both clients are built from source with ./configure
The openib.org server is AFAIK 1.1.x

This problem prevents me from changing my client to 1.2.0.
Is this a known issue?

Thanks,
No.2 | | 1963 bytes | |
Hi!

Quoting r. Michael S. Tsirkin <mst (AT) mellanox (DOT) co.il>:
Subject: issue with svn 1.2.0 client against 1.1.x server

Hello!
With subversion client version 1.2.0

>time /usr/local/svn12/bin/svn list https://openib.org/svn
branches/
gen1/
gen2/
tags/
trunk/
0.395u 0.067s 1:53.62 0.3% 0+0k 0+0io 0pf+0w

While with client 1.1.4 against the same server I have:

>time svn list https://openib.org/svn
branches/
gen1/
gen2/
tags/
trunk/
0.019u 0.007s 0:04.29 0.2% 0+0k 0+0io 11pf+0w

So subversion 1.2.0 is taking 2 minutes where 1.1.4 takes 4 seconds
to contact the same remote repository.

This observation was independently confirmed by another svn user.
Running strace on versions 1.2.0 and 1.1.4, I see that subversion 1.2.0
does around 17000 system calls, while 1.1.4 only around 750.
Further, most of the time while client 1.2.0 appears stuck,
it is actually performing read/write/select calls, presumably from the socket.

Both clients are built from source with ./configure
The openib.org server is AFAIK 1.1.x

This problem prevents me from changing my client to 1.2.0.
Is this a known issue?

I just re-tested with subversion 1.2.1 and I still have this issue.

/tmp>/usr/local/svn121/bin/svn
svn, version 1.2.1 (r15230)
compiled Jul 10 2005, 20:35:28
[]

/tmp>time /usr/local/svn121/bin/svn list https://openib.org/svn
[]
1.550u 0.130s 2:20.21 1.1% 0+0k 0+0io 832pf+0w

/tmp>svn
svn, version 1.1.4 (r13838)
compiled May 2 2005, 15:50:29
[]

/tmp>time svn list https://openib.org/svn
[]

0.070u 0.020s 0:04.28 2.1% 0+0k 0+0io 818pf+0w

openib.org subversion server is world-readable, so its easy to test.
Please contact me off-list if you have questions - I'm not subscribed.
No.3 | | 691 bytes | |
Jul 10, 2005, at 12:49 PM, Michael S. Tsirkin wrote:

openib.org subversion server is world-readable, so its easy to test.
Please contact me off-list if you have questions - I'm not subscribed.

What version of svn is openib.org running?

Is it possible to change openib.org to allow plain http:// access,
not just https://? Because of the SSL requirement, I can't analyze
the network trace. (It's anonymous access anyway, so I'm not sure
why encryption is being forced.)

To unsubscribe, e-mail: dev-unsubscribe (AT) subversion (DOT) tigris.org
For additional commands, e-mail: dev-help (AT) subversion (DOT) tigris.org
No.4 | | 1047 bytes | |
Jul 10, 2005, at 1:47 PM, Ben Collins-Sussman wrote:

Jul 10, 2005, at 12:49 PM, Michael S. Tsirkin wrote:
>
>>
>openib.org subversion server is world-readable, so its easy to test.
>Please contact me off-list if you have questions - I'm not
>subscribed.
>>
>
What version of svn is openib.org running?

The reason I ask is: the svn 1.2 client now uses 'peg revisions'
when peforming an 'ls' command, which means unlike a 1.1 client, it
asks the server to search through history. If the server is running
svn 1.2.0, this may be an example of the server-side history-
searching performance bugs fixed in svn 1.2.1. You'd have to patch
the server up to 1.2.1 to see if it makes a difference.

To unsubscribe, e-mail: dev-unsubscribe (AT) subversion (DOT) tigris.org
For additional commands, e-mail: dev-help (AT) subversion (DOT) tigris.org
No.5 | | 988 bytes | |
Quoting r. Ben Collins-Sussman <sussman (AT) collab (DOT) net>:
Subject: Re: svn 1.2.1 client x20 slower than 1.1.x

Jul 10, 2005, at 12:49 PM, Michael S. Tsirkin wrote:
>
>openib.org subversion server is world-readable, so its easy to test.
>Please contact me off-list if you have questions - I'm not subscribed.

What version of svn is openib.org running?

w3m https://openib.org/svn gets me

Powered by Subversion version 1.0.3 (r9775).

Is it possible to change openib.org to allow plain http:// access,
not just https://? Because of the SSL requirement, I can't analyze
the network trace. (It's anonymous access anyway, so I'm not sure
why encryption is being forced.)

Sorry, I'm not the webmaster, so I have no way to do that. Here's a link

http://openib.org/contact.html

Maybe subversion can be set up to print out activity on top of SSL?
No.6 | | 1527 bytes | |
K, this is a genuine bug, thanks for pointing it out again Michael
(and sorry we didn't fix it earlier!) I was able to reproduce the
bug by running the svn 1.2 client in gdb against the openib.org
server and watching what it was doing.

In short:

The svn 1.2 client is requesting *all* logs from the server, before
it ever issues an 'ls' request. And there's no reason for it do be
doing that. That explains the 2 minute delay.

In long:

As I suspected, the 1.2 svn client is doing history-tracing during an
'svn ls', when an svn 1.1 client is not.

Because openib.org is running svn 1.0, the client's attempt to call
svn_ra_get_locations() fails, so it falls back to running
slow_locations(), which ultimately calls svn_ra_get_logs().

In gdb, I can see that slow_locations() is called with arguments of
{start_revnum, end_revnum, peg_revnum} all set to 2829 (which is
HEAD). Yet somehow, at libsvn_client/ra.c:660, we call
svn_ra_get_logs() on a range of "youngest through r1". And the
revision 1 is hard-coded!

So instead of noticing, "hey, all the revisions are identical" and
doing zero history requests, the current logic says to
unconditionally fetch all logs back to r1. I have no idea why this
is. Can somebody else comment?

To unsubscribe, e-mail: dev-unsubscribe (AT) subversion (DOT) tigris.org
For additional commands, e-mail: dev-help (AT) subversion (DOT) tigris.org
No.7 | | 556 bytes | |
Sun, 2005-07-10 at 15:50 -0500, Ben Collins-Sussman wrote:
So instead of noticing, "hey, all the revisions are identical" and
doing zero history requests, the current logic says to
unconditionally fetch all logs back to r1. I have no idea why this
is. Can somebody else comment?

Perhaps you could try asking the author of that code? I suggest
starting at r9355. :)

To unsubscribe, e-mail: dev-unsubscribe (AT) subversion (DOT) tigris.org
For additional commands, e-mail: dev-help (AT) subversion (DOT) tigris.org
No.8 | | 711 bytes | |
Sun, 2005-07-10 at 15:50 -0500, Ben Collins-Sussman wrote:
The svn 1.2 client is requesting *all* logs from the server, before
it ever issues an 'ls' request. And there's no reason for it do be
doing that. That explains the 2 minute delay.

I fixed this in r15315, and nominated it for 1.2.2.

I'm not sure if this was made clear to Michael or not: although this
performance problem arises from an inefficiency in the client code,
upgrading the server to 1.1.x or higher will resolve the issue.

To unsubscribe, e-mail: dev-unsubscribe (AT) subversion (DOT) tigris.org
For additional commands, e-mail: dev-help (AT) subversion (DOT) tigris.org

Re: issue with svn 1.2.0 client against 1.1.x server

QUESTION ON "Development"

EMSDN.COM

It's a question for 'issue with svn 1.2.0 client against 1.1.x server' under 'Development'

Development

issue with svn 1.2.0 client against 1.1.x server