Sybase NNTP forums - End Of Life (EOL)

The NNTP forums from Sybase - forums.sybase.com - are now closed.

All new questions should be directed to the appropriate forum at the SAP Community Network (SCN).

Individual products have links to the respective forums on SCN, or you can go to SCN and search for your product in the search box (upper right corner) to find your specific developer center.

Sybase 11 NT server becoming unresponsive

5 posts in Windows NT Last posting was on 1997-10-31 01:42:12.0Z
David Mullen Posted on 1997-10-22 21:48:51.0Z
From: dave.mullen@emjay.com (David Mullen)
Subject: Sybase 11 NT server becoming unresponsive
Date: Wed, 22 Oct 1997 21:48:51 GMT
Organization: Emjay Corporation
Message-ID: <344f7465.81402493@forums.powersoft.com>
X-Newsreader: Forte Agent 1.5/32.451
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Newsgroups: sybase.public.sqlserver.nt
Lines: 36
Path: forums-1-dub!forums-master.sybase.com!forums.powersoft.com
Xref: forums-1-dub sybase.public.sqlserver.nt:5463
Article PK: 1081541

We are new to Sybase (developing our first application), and have
experienced the following problem:

When testing one of the functions in our new app, the client app would
suddenly seem to hang. Windows 95 would list the client app as "Not
responding". Eventually (within a few hours), *all* Sybase clients,
including WISQL, isql, etc. would all hang this way when trying to
connect to the Sybase server. Investigation showed that SQL Server
(and NT) was actually idle, apparently (see below) because the server
wasn't hearing the connect requests. So the server wasn't actually
hung or locked up, it was just unresponsive. Shutting down and
restarting SQL Server would resolve the problem.

This has been happening with version 11.0.2 on NT 4.0 (service pack
1). Sybase tech support says it's a bug in the netlib layer, and had
me apply SRW 7305, so that we're now at version 11.0.3. So far we've
been running for about a week without a recurrence. The problem is, my
developers and management are unsettled by the fact that Sybase have
been unable to tell us how we can be sure the problem is fixed, as in
"How can I see if the problem is occurring?" "You can't." "How can I
tell if the problem is fixed?" "The server keeps running."

So my question is, has anyone else had experience with this problem?
Can you tell me more about the problem? What was your solution? Are
you convinced it's fixed?

TIA!


Dick Brooks Posted on 1997-10-24 20:00:07.0Z
From: "Dick Brooks" <dick_b@ceo.sts-systems.ca>
References: <344f7465.81402493@forums.powersoft.com>
Subject: Re: Sybase 11 NT server becoming unresponsive
Date: Fri, 24 Oct 1997 16:00:07 -0400
Lines: 32
X-Newsreader: Microsoft Outlook Express 4.71.1712.3
X-MimeOle: Produced By Microsoft MimeOLE V4.71.1712.3
Message-ID: <mBUQk5K48GA.204@forums.powersoft.com>
Newsgroups: sybase.public.sqlserver.nt
Path: forums-1-dub!forums-master.sybase.com!forums.powersoft.com
Xref: forums-1-dub sybase.public.sqlserver.nt:5458
Article PK: 1081537

We've seen the same problem over several months running NT 4.0 sp3 on an SMP
Pentium box. Had it with 11.0.2.2 EBF6881, still have it with 11.0.3
EBF7305.

Can't do sp_who to investigate because you can't log in; the Sybase server
is still up but no longer accepting user connections; clients attempting
to connect hang until they timeout. The server errorlog seems to show a lot
of 1608 errors at this point, and eventually starts with 1601's (out of
connections) so it seems the connection attempt is being partially processed
to the extent of consuming a connection structure in the server.

A case logged with Tech Support has gone nowhere: at first they suggested
upgrading client and server to current EBF's, which didn't cure the problem.
Then it was probably network errors (even though several hundred Windows
clients and a few dozen Unix and NT servers on our network show no other
sign of network problems) and they basically refused to go any further until
we acquired a Sniffer and proved to them that our network is clean, and
forced users to disconnect in a disciplined way (yeah, right!). We do get a
few random 1608 and unexepected disconnects in the errorlog, but I consider
that normal.

We were getting the problem intermittently on our heavily used SMP
development box, generally about once a week. So far I haven't seen it on a
uniprocessor box, nor on a couple of SMP systems recently installed at
customer sites. The server is running only TCP/IP and Winsock, although
there's also Netbeui and IPX traffic on the wire. For now the "fix" has been
to bounce the server in our overnight script.

So no, I'm not convinced that it's been fixed.


David Mullen Posted on 1997-10-29 18:07:20.0Z
From: dave.mullen@emjay.com (David Mullen)
Subject: Re: Sybase 11 NT server becoming unresponsive
Date: Wed, 29 Oct 1997 18:07:20 GMT
Organization: Emjay Corporation
Message-ID: <345779b0.13123606@forums.powersoft.com>
References: <344f7465.81402493@forums.powersoft.com> <mBUQk5K48GA.204@forums.powersoft.com>
X-Newsreader: Forte Agent 1.5/32.451
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Newsgroups: sybase.public.sqlserver.nt
Lines: 80
Path: forums-1-dub!forums-master.sybase.com!forums.powersoft.com
Xref: forums-1-dub sybase.public.sqlserver.nt:5452
Article PK: 1081533

On Fri, 24 Oct 1997 14:13:34 -0400, Dick Brooks

<DBrooks@STSSystems.com> wrote:

>We've seen the same problem over several months running NT 4.0 sp3 on an
>SMP Pentium box. Had it with 11.0.2.2 EBF6881, still have it with 11.0.3
>EBF7305.
>
>Can't do sp_who to investigate because you can't log in; the Sybase
>server is still up but no longer accepting user connections; clients
>attempting to connect hang until they timeout. The server errorlog seems
>to show a lot of 1608 errors at this point, and eventually starts with
>1601's (out of connections) so it seems the connection attempt is being
>partially processed to the extent of consuming a connection structure in
>the server.

Yes, those are exactly the symptoms. We haven't had the 1601s, because
once the server locks up, people don't keep trying to connect. Ours is
still a development box; we don't go to production for another month.

>
>A case logged with Tech Support has gone nowhere: at first they
>suggested upgrading client and server to current EBF's, which didn't
>cure the problem. Then it was probably network errors (even though
>several hundred Windows clients and a few dozen Unix and NT servers on
>our network show no other sign of network problems) and they basically
>refused to go any further until we acquired a Sniffer and proved to them
>that our network is clean, and forced users to disconnect in a
>disciplined way (yeah, right!). We do get a few random 1608 and
>unexepected disconnects in the errorlog, but I consider that normal.

Do you get nl_read_defer errors in the log? All of our 1608s were
preceeded by nl_read_defers. My concern is that, while applying 7305
seems to have fixed it (at least, the server hasn't locked up again),
I still see nl_read_defers, which makes me suspicious that the
problem's still there. This is a Compaq Proliant 2500, single
processor Pentium Pro 200, 256Mb RAM, all alone on a 100 Mb/sec
network segment (full duplex), getting very light traffic from four
developers, nothing besides Sybase on the box. We've never had more
than twenty or so spids at once.

When we were getting the problem, it would take about a week to
develop.

>
>We were getting the problem intermittently on our heavily used SMP
>development box, generally about once a week. So far I haven't seen it
>on a uniprocessor box, nor on a couple of SMP systems recently installed
>at customer sites. The server is running only TCP/IP and Winsock,
>although there's also Netbeui and IPX traffic on the wire. For now the
>"fix" has been to bounce the server in our overnight script which has
>reduced the incidence to "not yet".
>
>So no, I'm not convinced that it's been fixed.
>
>It sounds like you're seeing the same problem. How similar is your
>setup? SMP with multiple Sybase engines? Winsock?

Uniprocessor, single SQL Server. Clients are Win95 via IPX. Server
also listens to named pipes (used by clients on NT machine, e.g. isql,
SQL Central) and TCP/IP (unused at present).

My main concern is that Sybase haven't been able to tell me why it was
happening, so my boss & developers aren't confident that it's fixed.


Philip Stafford-Jones Posted on 1997-10-31 01:42:12.0Z
Message-ID: <34593774.1BC7@sybase.com>
Date: Fri, 31 Oct 1997 12:42:12 +1100
From: Philip Stafford-Jones <psj@sybase.com>
Reply-To: psj@sybase.com
Organization: Sybase, pse server europe
X-Mailer: Mozilla 3.01 (WinNT; U)
MIME-Version: 1.0
CC: psj
Subject: Re: Sybase 11 NT server becoming unresponsive
References: <344f7465.81402493@forums.powersoft.com> <mBUQk5K48GA.204@forums.powersoft.com> <345779b0.13123606@forums.powersoft.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Newsgroups: sybase.public.sqlserver.nt
Lines: 35
Path: forums-1-dub!forums-master.sybase.com!forums.powersoft.com
Xref: forums-1-dub sybase.public.sqlserver.nt:5445
Article PK: 1081523

These hanging problems have been fixed by bugid's 131195 and 125610,
which are included in the 11.0.3 release EBF 7305 and well as
EBF 7407 (11.0.2.2 base + bugid's 131195, 125610 and 71859).

If you are encoutering these hanging problems please upgrade to
11.0.3 EBF 7305.

One other customer (Dick Brooks) is actually suffering a different
issue in that his client programs are abnormally disconnecting (perhaps
due to being turned off) this results in 1608 error messages in the
error log. Unfortunately the SQL Server is not informed about these
disconnections and they consume user connections, eventually the server
runs out of connections and issue the 1601 error message.

This will occur when you have many abnormal client disconnection
within a two hour period.

The solution is to reduce the TCP_KEEPALIVE period within the NT
Operating System (Refer to NT Resource Guide for details of how to
do this). This is the period at which the operating system probes
for dead connections and by default is several hours. Reducing this
to say 10 minutes will mean that abnormal client disconnects will
be spotted by the NT operating system within 10 minutes which will
then inform the Sybase SQL server which will remove the connection.

I hope this answers your queries,

Philip Stafford-Jones
Senior Product Specialist Engineer (Server Group)


Luis Oliveira Posted on 1997-10-27 19:45:01.0Z
From: "Luis Oliveira" <tsc@softcons.com>
References: <344f7465.81402493@forums.powersoft.com> <mBUQk5K48GA.204@forums.powersoft.com>
Subject: Re: Sybase 11 NT server becoming unresponsive
Date: Mon, 27 Oct 1997 13:45:01 -0600
Lines: 14
X-Newsreader: Microsoft Outlook Express 4.71.1008.3
X-MimeOle: Produced By Microsoft MimeOLE Engine V4.71.1008.3
Message-ID: <aEanigw48GA.151@forums.powersoft.com>
Newsgroups: sybase.public.sqlserver.nt
Path: forums-1-dub!forums-master.sybase.com!forums.powersoft.com
Xref: forums-1-dub sybase.public.sqlserver.nt:5456
Article PK: 1081534

We had a similar problem, and according to Technical support this was due
to a bug and it would only happen in multi-processor machines, where we
would allow SQL Server to use more than one CPU. The problem seems to have
been fixed after we installed EBF 7407.

Luis Oliveira
Landry's Seafood Restaurants
lco@softcons.com