What does send sleep mean in Sybase central(processes)
shankar kamath
I am using Delphi4.0 with sybase(11.5). In my application, users search
for records very offten( generally a simple select statement with where
caluse) and very frequently user's machine freeze. Sybase central
processes shows status as
'send sleep'. command :SELECT
Appreciate your helps
Eric McGrane
send/recv sleep indicate that a client has been put into the sleep queue and
is either waiting for a tds packet from the server, send sleep, or the
server is waiting for a packet from the client, recv sleep.
So, if you see many clients in send sleep, this indicates that they are
sleeping waiting for the server to send them a packet.
E
shankar kamath
How can we resolve this send sleep condition ? If I view from windows NT
event viewer, it
says "120191 : Can not send. host process disconnected"
Please help
Shankar
Eric McGrane
The fact that you are seeing messages in the log like can not send host process
disconnected means that the clients "at the other end of the line" are no longer
there. I am guessing that your clients are pc clients and they are getting
impatient and rebooting. The result is that ASE has no idea the client has gone
away until it actually tries to communicate with the client, when it does you get
the 'host process disconnected message'.
I would first educate the users to not be so impatient and explain the
ramifications of rebooting. If they are experiencing performance problems those
problems need to be reported to the dba and investigated to understood. Only then
will this be resolved.
I am sorry I can't give you the real answer you are looking for. I can only point
you in the right direction at this time.
Dee Roghelia
September on ASE 11.5 on NT SP4. A stored procedure that is frequently run,
locks while the user processes sit in a "send sleep" state; in our case, the
data is never returned (we have let it go for over 1 hour on a stored
procedure that should return results in 5 seconds). And no matter how much
we have tried to train the users not to reboot, they do anyway. This seems
to make the server even more bogged down because it is trying to clean up
these processes once it determines that they have been disconnected. The
problem is also intermittent. It seems to get worse as more users log on to
the system.
We resorted to calling Sybase Professional Services in November. Once at my
client's site, a couple of things the Consultant tried was first to increase
the amount of memory for the OS. Then we added an entry to the NT registry
for the Keep Alive setting, which will help detect client disconnects
(Document ID 611 on Sybase's site). The default for this type of checking
on most platforms is 2 hours. We also upgraded ASE to 11.5.1 and then
applied SWR 8608, which includes a fix for Bug 187326: "Unexpected client
disconnect (eg due to network error or client dying) can leave 'orphan'
process on ASE." These fixes did help reduce the problem but the Sybase
Consultant suggested we apply NT SP5, which deals with tcp/ip issues that
may be causing our problem. We are currently waiting for this request to be
approved by my client.
I hope this helps. Good Luck! This type of situation can be very
frustrating for you and your users!!!
Dee Roghelia
Eric McGrane <mcg...@sybase.com> wrote in message
news:387B38D5...@sybase.com...
Gary Estridge
frequent page faulting that was occurring on the NT server, it really didn't
have anything to do with the disconnects that were occurring. We gave the
OS more memory to work with which stopped all the page faulting that was
going on. Page faulting can seriously degrade server performance. All of
the other items you mentioned are the things he should check out.
Gary Estridge
Sr. Consultant
Sybase Professional Services
"Dee Roghelia" <drog...@ciber.com> wrote in message
news:11zTmGEX$GA....@forums.sybase.com...