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.

Network Slowdowns ADS 9.1

4 posts in Networking Last posting was on 2011-10-28 20:08:57.0Z
Jerry Kogan Posted on 2011-10-27 18:23:29.0Z
Reply-To: "Jerry Kogan" <jkogan@cksoftware.com>
From: "Jerry Kogan" <jkogan@cksoftware.com>
Newsgroups: advantage.networking
Subject: Network Slowdowns ADS 9.1
Date: Thu, 27 Oct 2011 12:23:29 -0600
Lines: 43
Organization: CK Software, Inc
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2900.5931
X-RFC2646: Format=Flowed; Original
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.6109
NNTP-Posting-Host: 71.216.251.153
Message-ID: <4ea9a1a1$1@solutions.advantagedatabase.com>
X-Trace: 27 Oct 2011 11:23:29 -0700, 71.216.251.153
Path: solutions.advantagedatabase.com
Xref: solutions.advantagedatabase.com Advantage.Networking:967
Article PK: 1132376

We are encountering intermittent 1-3 minute slowdowns on a Windows 2003
Server with ADS 9.1. The server has a Quad-Core processor and is generally
using only a tiny fraction of its CPU and the network bandwidth. The
slowdowns appears to freeze applications that are trying to pull information
from the network. Whenever they occur and are reported they are at roughly
two hour intervals. One day they might be reported at 9:41 and at 3:41.
The next at 9:43, 11:43 and 5:43. It's possible that something is
happening every two hours, but that users are not always noticing it. Every
day the times that these freezes might occur seems to advance about 2
minutes. We have been monitoring these times for several weeks and they
continue to follow this regular pattern. Time of occurences every day are
two to three minutes later, and times within the day are multiples of two
hours apart. I have checked the ADS_ERR table and do not find errors
corresponding to the times of these freezes. System error logs also do not
show do not show entries corresponding to these freeze up times.

During these freezes users can switch to other tasks and these will procede
normally as long as they do not try to access files on the network. We have
monitored the server during some of these freezes and are starting to see
the following. The overall % disk time hits 100% for a minute or more. ADS
Page Faults also skyrocket. The system administrator feels that it is the
ADS Page Faults which are causing the disk activity. The disk activity
slows the system so much that the performance monitor quits recording and
most actual data points from the worst of the freeze-up are missing. The %
processor time also increases. To me the % of ADS Page Faults seems pretty
high for several minutes on either side of the slowdown. The ADS %
processor time also increases.
I have examined user logs I have set up within my Delphi program and do not
see anything that users are doing at these times that is different than at
other times. So I don't think it's something specific to users. Our system
administrator feels strongly that ADS is doing something internally at
roughly two hours intervals that is bringing the network to a crawl. He
says that he has made sure that the Kaspersky Anti-Virus program is not
running scans at two hour intervals, and is generally not running network
intensive stuff during working hours. He has seen nothing in any other logs
that would point to other applications or processes.
So my question is whether there is some validity to his suspicion. Is there
anything that ADS does internally at two hour intervals. Is so, what is it,
and what can we do so it doesn't bog down the network so severely at those
times. If not, I'd appreciate any other suggestions on how we might be able
to narrow down the source.


Mark Wilkins Posted on 2011-10-27 22:48:27.0Z
From: "Mark Wilkins" <a@b.c>
Newsgroups: advantage.networking
References: <4ea9a1a1$1@solutions.advantagedatabase.com>
In-Reply-To: <4ea9a1a1$1@solutions.advantagedatabase.com>
Subject: Re: Network Slowdowns ADS 9.1
Date: Thu, 27 Oct 2011 16:48:27 -0600
Lines: 1
Organization: Sybase
MIME-Version: 1.0
Content-Type: text/plain; format=flowed; charset="iso-8859-1"; reply-type=response
Content-Transfer-Encoding: 7bit
X-Priority: 3
X-MSMail-Priority: Normal
Importance: Normal
X-Newsreader: Microsoft Windows Live Mail 14.0.8089.726
X-MimeOLE: Produced By Microsoft MimeOLE V14.0.8089.726
NNTP-Posting-Host: 10.6.199.122
Message-ID: <4ea9dfb9$1@solutions.advantagedatabase.com>
X-Trace: 27 Oct 2011 15:48:25 -0700, 10.6.199.122
Path: solutions.advantagedatabase.com
Xref: solutions.advantagedatabase.com Advantage.Networking:968
Article PK: 1132377

Hi Jerry,

One thing that came to my mind is a process in ADS that flushes file buffers
on a timed basis. I believe the default is 10 minutes. You might look in
the ADS configuration for a config entry named FLUSH_FREQUENCY. The value
is in milliseconds. Is it perhaps set to 7200000? I don't know if it could
produce the behavior you are seeing, but when the interval passes, ADS spins
through open files and calls a Windows API to force writes to disk.

Aside from that, I'm not aware of anything that could happen on that type of
interval. There are a number of things that are processed by threads on a
regular basis, but they run every few seconds typically. And the results of
those operations are highly dependent on the state. For example, a sweep
thread monitors the cache system and can dump blocks from the cache to disk,
but the actual behavior depends on how much memory is in use by ADS, by
other applications, the activity in ADS, etc.

Mark Wilkins
Advantage R&D

"Jerry Kogan" <jkogan@cksoftware.com> wrote in message
news:4ea9a1a1$1@solutions.advantagedatabase.com...
> We are encountering intermittent 1-3 minute slowdowns on a Windows 2003
> Server with ADS 9.1. The server has a Quad-Core processor and is generally
> using only a tiny fraction of its CPU and the network bandwidth. The
> slowdowns appears to freeze applications that are trying to pull
> information from the network. Whenever they occur and are reported they
> are at roughly two hour intervals. One day they might be reported at 9:41
> and at 3:41. The next at 9:43, 11:43 and 5:43. It's possible that
> something is happening every two hours, but that users are not always
> noticing it. Every day the times that these freezes might occur seems to
> advance about 2 minutes. We have been monitoring these times for several
> weeks and they continue to follow this regular pattern. Time of occurences
> every day are two to three minutes later, and times within the day are
> multiples of two hours apart. I have checked the ADS_ERR table and do not
> find errors corresponding to the times of these freezes. System error logs
> also do not show do not show entries corresponding to these freeze up
> times.
>


Jerry Kogan Posted on 2011-10-28 06:38:10.0Z
Reply-To: "Jerry Kogan" <jkogan@cksoftware.com>
From: "Jerry Kogan" <jkogan@cksoftware.com>
Newsgroups: advantage.networking
References: <4ea9a1a1$1@solutions.advantagedatabase.com> <4ea9dfb9$1@solutions.advantagedatabase.com>
Subject: Re: Network Slowdowns ADS 9.1
Date: Fri, 28 Oct 2011 00:38:10 -0600
Lines: 53
Organization: CK Software, Inc
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2900.5931
X-RFC2646: Format=Flowed; Response
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.6109
NNTP-Posting-Host: 71.216.251.153
Message-ID: <4eaa4dd6$1@solutions.advantagedatabase.com>
X-Trace: 27 Oct 2011 23:38:14 -0700, 71.216.251.153
Path: solutions.advantagedatabase.com
Xref: solutions.advantagedatabase.com Advantage.Networking:969
Article PK: 1132379

Mark,
I checked the registry and FLUSH_FREQUENCY is set at 600000.
Is there any way to get a log file to show when some of this stuff is
happening?
Also I don't see any kind of ASSERT.LOG. Shouldn't it be in the same folder
as the ADS_ERR.ADT table?

"Mark Wilkins" <a@b.c> wrote in message
news:4ea9dfb9$1@solutions.advantagedatabase.com...
> Hi Jerry,
>
> One thing that came to my mind is a process in ADS that flushes file
> buffers on a timed basis. I believe the default is 10 minutes. You might
> look in the ADS configuration for a config entry named FLUSH_FREQUENCY.
> The value is in milliseconds. Is it perhaps set to 7200000? I don't know
> if it could produce the behavior you are seeing, but when the interval
> passes, ADS spins through open files and calls a Windows API to force
> writes to disk.
>
> Aside from that, I'm not aware of anything that could happen on that type
> of interval. There are a number of things that are processed by threads
> on a regular basis, but they run every few seconds typically. And the
> results of those operations are highly dependent on the state. For
> example, a sweep thread monitors the cache system and can dump blocks from
> the cache to disk, but the actual behavior depends on how much memory is
> in use by ADS, by other applications, the activity in ADS, etc.
>
> Mark Wilkins
> Advantage R&D
>
> "Jerry Kogan" <jkogan@cksoftware.com> wrote in message
> news:4ea9a1a1$1@solutions.advantagedatabase.com...
>> We are encountering intermittent 1-3 minute slowdowns on a Windows 2003
>> Server with ADS 9.1. The server has a Quad-Core processor and is
>> generally using only a tiny fraction of its CPU and the network
>> bandwidth. The slowdowns appears to freeze applications that are trying
>> to pull information from the network. Whenever they occur and are
>> reported they are at roughly two hour intervals. One day they might be
>> reported at 9:41 and at 3:41. The next at 9:43, 11:43 and 5:43. It's
>> possible that something is happening every two hours, but that users are
>> not always noticing it. Every day the times that these freezes might
>> occur seems to advance about 2 minutes. We have been monitoring these
>> times for several weeks and they continue to follow this regular pattern.
>> Time of occurences every day are two to three minutes later, and times
>> within the day are multiples of two hours apart. I have checked the
>> ADS_ERR table and do not find errors corresponding to the times of these
>> freezes. System error logs also do not show do not show entries
>> corresponding to these freeze up times.
>>
>
>


Mark Wilkins Posted on 2011-10-28 20:08:57.0Z
From: "Mark Wilkins" <a@b.c>
Newsgroups: advantage.networking
References: <4ea9a1a1$1@solutions.advantagedatabase.com> <4ea9dfb9$1@solutions.advantagedatabase.com> <4eaa4dd6$1@solutions.advantagedatabase.com>
In-Reply-To: <4eaa4dd6$1@solutions.advantagedatabase.com>
Subject: Re: Network Slowdowns ADS 9.1
Date: Fri, 28 Oct 2011 14:08:57 -0600
Lines: 3
Organization: Sybase
MIME-Version: 1.0
Content-Type: text/plain; format=flowed; charset="iso-8859-1"; reply-type=response
Content-Transfer-Encoding: 7bit
X-Priority: 3
X-MSMail-Priority: Normal
Importance: Normal
X-Newsreader: Microsoft Windows Live Mail 14.0.8089.726
X-MimeOLE: Produced By Microsoft MimeOLE V14.0.8089.726
NNTP-Posting-Host: 10.6.199.122
Message-ID: <4eab0bd4$1@solutions.advantagedatabase.com>
X-Trace: 28 Oct 2011 13:08:52 -0700, 10.6.199.122
Path: solutions.advantagedatabase.com
Xref: solutions.advantagedatabase.com Advantage.Networking:970
Article PK: 1132380

Hi Jerry,

If there were an assertion failure, then there would be an assert.log file
in the same location as the ads_err.* files. There would also like be a
file named adsdump-<digits>-<moredigits>.dmp.gz. The dump file would
indicate some kind of internal error situation that caused it to write out
the process information. Typically, though, I think those are written at
most once per 24 hours (well - ideally never but if an error situation his
hit repeatedly it is throttled to once per day). If the process (memory)
size of ADS is very large, it can take several minutes to write the dump
file and it would basically stop everything. However, it would not affect
network traffic, which it sounded like was happening.

The Windows event log will sometimes contain information too, although it
typically would correspond to dump files being created. But you might look
there to see if anything is happening during those times. Aside from those
things (error logs, asssert log, dump files, and event log) I am not
remembering any other obvious things to check.

Mark Wilkins
Advantage R&D

"Jerry Kogan" <jkogan@cksoftware.com> wrote in message
news:4eaa4dd6$1@solutions.advantagedatabase.com...
> Mark,
> I checked the registry and FLUSH_FREQUENCY is set at 600000.
> Is there any way to get a log file to show when some of this stuff is
> happening?
> Also I don't see any kind of ASSERT.LOG. Shouldn't it be in the same
> folder as the ADS_ERR.ADT table?
>
> "Mark Wilkins" <a@b.c> wrote in message
> news:4ea9dfb9$1@solutions.advantagedatabase.com...
>> Hi Jerry,
>>
>> One thing that came to my mind is a process in ADS that flushes file
>> buffers on a timed basis. I believe the default is 10 minutes. You
>> might look in the ADS configuration for a config entry named
>> FLUSH_FREQUENCY. The value is in milliseconds. Is it perhaps set to
>> 7200000? I don't know if it could produce the behavior you are seeing,
>> but when the interval passes, ADS spins through open files and calls a
>> Windows API to force writes to disk.
>>
>> Aside from that, I'm not aware of anything that could happen on that type
>> of interval. There are a number of things that are processed by threads
>> on a regular basis, but they run every few seconds typically. And the
>> results of those operations are highly dependent on the state. For
>> example, a sweep thread monitors the cache system and can dump blocks
>> from the cache to disk, but the actual behavior depends on how much
>> memory is in use by ADS, by other applications, the activity in ADS, etc.
>>
>> Mark Wilkins
>> Advantage R&D
>>
>> "Jerry Kogan" <jkogan@cksoftware.com> wrote in message
>> news:4ea9a1a1$1@solutions.advantagedatabase.com...
>>> We are encountering intermittent 1-3 minute slowdowns on a Windows 2003
>>> Server with ADS 9.1. The server has a Quad-Core processor and is
>>> generally using only a tiny fraction of its CPU and the network
>>> bandwidth. The slowdowns appears to freeze applications that are trying
>>> to pull information from the network. Whenever they occur and are
>>> reported they are at roughly two hour intervals. One day they might be
>>> reported at 9:41 and at 3:41. The next at 9:43, 11:43 and 5:43. It's
>>> possible that something is happening every two hours, but that users are
>>> not always noticing it. Every day the times that these freezes might
>>> occur seems to advance about 2 minutes. We have been monitoring these
>>> times for several weeks and they continue to follow this regular
>>> pattern. Time of occurences every day are two to three minutes later,
>>> and times within the day are multiples of two hours apart. I have
>>> checked the ADS_ERR table and do not find errors corresponding to the
>>> times of these freezes. System error logs also do not show do not show
>>> entries corresponding to these freeze up times.
>>>
>>
>>
>
>