[p4] Anyone experienced these perforce problems? - Jacob Ritorto - com.perforce.perforce-user

9 messages in com.perforce.perforce-user[p4] Anyone experienced these perforc...

From	Sent On	Attachments
Rajesh Vaidheeswarran	26 Jun 2001 11:33
Richard Geiger	26 Jun 2001 12:29
Jacob Ritorto	26 Jun 2001 12:32
Chuck Karish	26 Jun 2001 14:46
Stephen Vance	26 Jun 2001 19:27
Stephen Vance	26 Jun 2001 19:56
Dave Lewis	26 Jun 2001 20:29
Rajesh Vaidheeswarran	27 Jun 2001 11:01
Chuck Karish	27 Jun 2001 12:15

Subject:	[p4] Anyone experienced these perforce problems?
From:	Jacob Ritorto (jrit...@fore.com)
Date:	06/26/2001 12:32:14 PM
List:	com.perforce.perforce-user

Rajesh, I can't really address your FreeBSD and Linux questions, and I don't know what to tell you about the packet loss in the net. But as a datapoint, I'll mention that I've been running p4d on a single SPARC Ultra2 with storage on a NetApp, significantly more users than you and a (ludicrous) 27-gig depot for over two years without a serious failure. In fact I'm currently running through a recovery drill because the reliability has been so good it's unnerving. ISTR chatter about the nfs locking stuff in Redhat not being up to snuff for this sort of application when RH6.2 was out, but I forget the details. Of course they may have mended it in the 7.x releases. Nonetheless, I'm pretty satisfied with my scenario.

[p4]:/> uptime 2:57pm up 319 day(s), 6:56, 1 user, load average: 0.05, 0.05, 0.07 [p4]:/> uname -a SunOS p4 5.6 Generic_105181-11 sun4u sparc SUNW,Ultra-2 [p4]:/>

regards, --jake

On Tue, 26 Jun 2001, Rajesh Vaidheeswarran wrote:

Hi,

We use Perforce here with about 30 users, mostly on Windows workstations. Server is a solaris 5.8, running on an Ultra 80. The storage is on a Netapp server, 6.1r1.

We have been encountering *very long* delays from the client side for many p4 commands .. a p4 edit which typically returns in under a second returns at times after 10 minutes! Most often it times out..

p4 syncs fail sometimes. p4 edits time out..

Since I've used perforce in a previous job, I know for a fact that this is not a perforce server issue, since we had over 250 engineers in multiple sites doing p4 operations (sometimes across distributed depots.)

A snoop at the server end shows the server arbitrarily (or so it seems,)chooses not to ACK some packets sent to the P4PORT. The client sometimes times out and retransmits, and the dialog continues.. mostly, it ends with the client dying like this..

%p4 info Perforce client error: Connection with partner closed unexpectedly. TCP receive failed. read: socket: WSAECONNRESET

The server side shows some very interesting logs. Log of errors from this morning are listed below.

My question is:

Has anyone had these types of problems in a similar configuration with a sparc and a netapp.. I cannot confirm it, but some of our users say that they started noticing these problems since the netapp was added to the picture, replacing the local disk.

We are evaluating replacing the server with an intel box running either FreeBSD or RH/Linux. Are there any recommendations that any of you can make on whether we should consider one over the other (or if we should consider a third OS instead.) There is a distinct advantage for us to go with Linux because of the installed base of Linux servers. We have no FreeBSD machines here.

Thanks rv

Errors on the server

--------------------

Perforce server error: Date 2001/06/26 08:45:29: Operation: user-sync TCP send failed. write: socket: Broken pipe Can't invoke remote operation 'client-WriteFile'.