We are having an issue with the application pool being disabled due to
rapid-fail protection. My best guess is that a routine admin script which
is supposed to walk the metabase and copy the previous day's logs to an
appropriately named folder on an archive box for stats, is locking the
metabase. So far, I have been unsuccessful at correcting the problem.

Basic situation is a stand alone shared hosting box set with each
customer's sites being set up in an account specific application pool.
(Might be more than one virtual server per application pool depending on the
customer.) The application pools are set to shutdown worker processes after
being idle 20 minutes. Some of the sites are just not very busy so they do
shut down.

And it is those which seem to be having the application pool failure. The
W3svc log for every one that has failed shows no activity for an extended
time. The System log shows Event ID 1009 five times in less than a second,
("A process serving application pool 'xxxxxxxx' terminated unexpectedly. The
process id was '1604'. The process exit code was '0xffffffff'.") and is then
followed by event 1002 shutting down the application pool. It looks like a
spider, script kiddie or random visitor timed their request perfectly so the
scheduled task was finishing.

Last night, the Application pool failures were 9:08:26PM to 9:08:27PM. The
scheduled task running the admin script started at 9:07:00 PM and was logged
off at 9:08:30PM. Actually, it should have been done by 9:07:09PM -- that
is when UPHClean logged an entry it had done cleanup. The script is using
ADSI, nothing really strange, nothing that should be changing the metabase.
But from the evidence of the crashing application pool, seems to be locking
the metabase anyway.

[section of code where it starts finding the sites -- just in case I am not
explaining this script adequately]
Set objW3SVC = GetObject("IIS://LOCALHOST/W3SVC")
'Walk through the metabase, and pick out web servers only.
For Each objSITE in objW3SVC
If objSITE.class = "IIsWebServer" Then
servNum = objSITE.name
For each port in objSITE.ServerBindings


I could not see anything in the documentation indicating that this should be
a concern. Can someone help and point me to relevant docs? Is there a
read-only way to do this. Or is this a bug with ADSI and IIS 6? Are there
alternatives that would avoid the locking?



Thanks.

Mary Rafferty

Re: Application pool shutdown - scheduled admin script reading metabase by Bernard

Bernard
Sat Jun 18 22:51:04 CDT 2005

Not sure, but i don't think reading metabase will cause app pool to crash. I
would suggest you get DebugDiag and try to debug the app.

--
Regards,
Bernard Cheah
http://www.microsoft.com/iis/
http://www.iiswebcastseries.com/
http://www.msmvps.com/bernard/


"M. M. Rafferty" <mmr@vistagrande.com> wrote in message
news:uoauLYHdFHA.1448@TK2MSFTNGP14.phx.gbl...
> We are having an issue with the application pool being disabled due to
> rapid-fail protection. My best guess is that a routine admin script which
> is supposed to walk the metabase and copy the previous day's logs to an
> appropriately named folder on an archive box for stats, is locking the
> metabase. So far, I have been unsuccessful at correcting the problem.
>
> Basic situation is a stand alone shared hosting box set with each
> customer's sites being set up in an account specific application pool.
> (Might be more than one virtual server per application pool depending on
> the
> customer.) The application pools are set to shutdown worker processes
> after
> being idle 20 minutes. Some of the sites are just not very busy so they
> do
> shut down.
>
> And it is those which seem to be having the application pool failure. The
> W3svc log for every one that has failed shows no activity for an extended
> time. The System log shows Event ID 1009 five times in less than a
> second,
> ("A process serving application pool 'xxxxxxxx' terminated unexpectedly.
> The
> process id was '1604'. The process exit code was '0xffffffff'.") and is
> then
> followed by event 1002 shutting down the application pool. It looks like
> a
> spider, script kiddie or random visitor timed their request perfectly so
> the
> scheduled task was finishing.
>
> Last night, the Application pool failures were 9:08:26PM to 9:08:27PM.
> The
> scheduled task running the admin script started at 9:07:00 PM and was
> logged
> off at 9:08:30PM. Actually, it should have been done by 9:07:09PM -- that
> is when UPHClean logged an entry it had done cleanup. The script is
> using
> ADSI, nothing really strange, nothing that should be changing the
> metabase.
> But from the evidence of the crashing application pool, seems to be
> locking
> the metabase anyway.
>
> [section of code where it starts finding the sites -- just in case I am
> not
> explaining this script adequately]
> Set objW3SVC = GetObject("IIS://LOCALHOST/W3SVC")
> 'Walk through the metabase, and pick out web servers only.
> For Each objSITE in objW3SVC
> If objSITE.class = "IIsWebServer" Then
> servNum = objSITE.name
> For each port in objSITE.ServerBindings
>
>
> I could not see anything in the documentation indicating that this should
> be
> a concern. Can someone help and point me to relevant docs? Is there a
> read-only way to do this. Or is this a bug with ADSI and IIS 6? Are
> there
> alternatives that would avoid the locking?
>
>
>
> Thanks.
>
> Mary Rafferty
>
>



Re: Application pool shutdown - scheduled admin script reading metabase by M

M
Sun Jun 19 00:52:15 CDT 2005

Hi Bernard,

I would tend to agree that reading the metabase should not crash the app
pool. However, I have one instance where it happened to a site which is
temporarily set to redirect elsewhere -- nothing in the way of an IIS app to
debug as they still haven't uploaded the new version.

None of these sites have any issues at any other time. Only when they have
been shut down and the admin script has just executed does this happen.
There is nothing in the logs to provide a clue as to the specific request.
Note that most of these are very small sites and have little use of ASP or
ASP.NET except to supply the common header and footer elements.

It must be what the admin script is doing, or something else with IIS6 or
Win2K3 is broken.



mmr


"Bernard Cheah [MVP]" <qbernard@hotmail.com.discuss> wrote in message
news:e$ksfIIdFHA.3280@TK2MSFTNGP09.phx.gbl...
> Not sure, but i don't think reading metabase will cause app pool to crash.
I
> would suggest you get DebugDiag and try to debug the app.
>
> --
> Regards,
> Bernard Cheah
> http://www.microsoft.com/iis/
> http://www.iiswebcastseries.com/
> http://www.msmvps.com/bernard/
>
>
> "M. M. Rafferty" <mmr@vistagrande.com> wrote in message
> news:uoauLYHdFHA.1448@TK2MSFTNGP14.phx.gbl...
> > We are having an issue with the application pool being disabled due to
> > rapid-fail protection. My best guess is that a routine admin script
which
> > is supposed to walk the metabase and copy the previous day's logs to an
> > appropriately named folder on an archive box for stats, is locking the
> > metabase. So far, I have been unsuccessful at correcting the problem.
> >
> > Basic situation is a stand alone shared hosting box set with each
> > customer's sites being set up in an account specific application pool.
> > (Might be more than one virtual server per application pool depending on
> > the
> > customer.) The application pools are set to shutdown worker processes
> > after
> > being idle 20 minutes. Some of the sites are just not very busy so they
> > do
> > shut down.
> >
> > And it is those which seem to be having the application pool failure.
The
> > W3svc log for every one that has failed shows no activity for an
extended
> > time. The System log shows Event ID 1009 five times in less than a
> > second,
> > ("A process serving application pool 'xxxxxxxx' terminated unexpectedly.
> > The
> > process id was '1604'. The process exit code was '0xffffffff'.") and is
> > then
> > followed by event 1002 shutting down the application pool. It looks
like
> > a
> > spider, script kiddie or random visitor timed their request perfectly so
> > the
> > scheduled task was finishing.
> >
> > Last night, the Application pool failures were 9:08:26PM to 9:08:27PM.
> > The
> > scheduled task running the admin script started at 9:07:00 PM and was
> > logged
> > off at 9:08:30PM. Actually, it should have been done by 9:07:09PM --
that
> > is when UPHClean logged an entry it had done cleanup. The script is
> > using
> > ADSI, nothing really strange, nothing that should be changing the
> > metabase.
> > But from the evidence of the crashing application pool, seems to be
> > locking
> > the metabase anyway.
> >
> > [section of code where it starts finding the sites -- just in case I am
> > not
> > explaining this script adequately]
> > Set objW3SVC = GetObject("IIS://LOCALHOST/W3SVC")
> > 'Walk through the metabase, and pick out web servers only.
> > For Each objSITE in objW3SVC
> > If objSITE.class = "IIsWebServer" Then
> > servNum = objSITE.name
> > For each port in objSITE.ServerBindings
> >
> >
> > I could not see anything in the documentation indicating that this
should
> > be
> > a concern. Can someone help and point me to relevant docs? Is there a
> > read-only way to do this. Or is this a bug with ADSI and IIS 6? Are
> > there
> > alternatives that would avoid the locking?
> >
> >
> >
> > Thanks.
> >
> > Mary Rafferty
> >
> >
>
>



Re: Application pool shutdown - scheduled admin script reading metabase by Bernard

Bernard
Sun Jun 19 06:31:36 CDT 2005

Try debugdiag and hopefully it will able to find the root cause.

--
Regards,
Bernard Cheah
http://www.microsoft.com/iis/
http://www.iiswebcastseries.com/
http://www.msmvps.com/bernard/


"M. M. Rafferty" <mmr@vistagrande.com> wrote in message
news:O9xBWMJdFHA.1356@TK2MSFTNGP10.phx.gbl...
> Hi Bernard,
>
> I would tend to agree that reading the metabase should not crash the app
> pool. However, I have one instance where it happened to a site which is
> temporarily set to redirect elsewhere -- nothing in the way of an IIS app
> to
> debug as they still haven't uploaded the new version.
>
> None of these sites have any issues at any other time. Only when they
> have
> been shut down and the admin script has just executed does this happen.
> There is nothing in the logs to provide a clue as to the specific request.
> Note that most of these are very small sites and have little use of ASP or
> ASP.NET except to supply the common header and footer elements.
>
> It must be what the admin script is doing, or something else with IIS6 or
> Win2K3 is broken.
>
>
>
> mmr
>
>
> "Bernard Cheah [MVP]" <qbernard@hotmail.com.discuss> wrote in message
> news:e$ksfIIdFHA.3280@TK2MSFTNGP09.phx.gbl...
>> Not sure, but i don't think reading metabase will cause app pool to
>> crash.
> I
>> would suggest you get DebugDiag and try to debug the app.
>>
>> --
>> Regards,
>> Bernard Cheah
>> http://www.microsoft.com/iis/
>> http://www.iiswebcastseries.com/
>> http://www.msmvps.com/bernard/
>>
>>
>> "M. M. Rafferty" <mmr@vistagrande.com> wrote in message
>> news:uoauLYHdFHA.1448@TK2MSFTNGP14.phx.gbl...
>> > We are having an issue with the application pool being disabled due to
>> > rapid-fail protection. My best guess is that a routine admin script
> which
>> > is supposed to walk the metabase and copy the previous day's logs to an
>> > appropriately named folder on an archive box for stats, is locking the
>> > metabase. So far, I have been unsuccessful at correcting the problem.
>> >
>> > Basic situation is a stand alone shared hosting box set with each
>> > customer's sites being set up in an account specific application pool.
>> > (Might be more than one virtual server per application pool depending
>> > on
>> > the
>> > customer.) The application pools are set to shutdown worker processes
>> > after
>> > being idle 20 minutes. Some of the sites are just not very busy so
>> > they
>> > do
>> > shut down.
>> >
>> > And it is those which seem to be having the application pool failure.
> The
>> > W3svc log for every one that has failed shows no activity for an
> extended
>> > time. The System log shows Event ID 1009 five times in less than a
>> > second,
>> > ("A process serving application pool 'xxxxxxxx' terminated
>> > unexpectedly.
>> > The
>> > process id was '1604'. The process exit code was '0xffffffff'.") and is
>> > then
>> > followed by event 1002 shutting down the application pool. It looks
> like
>> > a
>> > spider, script kiddie or random visitor timed their request perfectly
>> > so
>> > the
>> > scheduled task was finishing.
>> >
>> > Last night, the Application pool failures were 9:08:26PM to 9:08:27PM.
>> > The
>> > scheduled task running the admin script started at 9:07:00 PM and was
>> > logged
>> > off at 9:08:30PM. Actually, it should have been done by 9:07:09PM --
> that
>> > is when UPHClean logged an entry it had done cleanup. The script is
>> > using
>> > ADSI, nothing really strange, nothing that should be changing the
>> > metabase.
>> > But from the evidence of the crashing application pool, seems to be
>> > locking
>> > the metabase anyway.
>> >
>> > [section of code where it starts finding the sites -- just in case I am
>> > not
>> > explaining this script adequately]
>> > Set objW3SVC = GetObject("IIS://LOCALHOST/W3SVC")
>> > 'Walk through the metabase, and pick out web servers only.
>> > For Each objSITE in objW3SVC
>> > If objSITE.class = "IIsWebServer" Then
>> > servNum = objSITE.name
>> > For each port in objSITE.ServerBindings
>> >
>> >
>> > I could not see anything in the documentation indicating that this
> should
>> > be
>> > a concern. Can someone help and point me to relevant docs? Is there a
>> > read-only way to do this. Or is this a bug with ADSI and IIS 6? Are
>> > there
>> > alternatives that would avoid the locking?
>> >
>> >
>> >
>> > Thanks.
>> >
>> > Mary Rafferty
>> >
>> >
>>
>>
>
>



Re: Application pool shutdown - scheduled admin script reading metabase by M

M
Mon Jun 20 23:54:40 CDT 2005

Hi Bernard,

I downloaded the tool and installed it on the server. Set one rule --
basically took the defaults for a crash for all IIS related processes. Then
I set it to active shortly before the copy logs script ran. At the
appointed time I tried to help create a crash by browsing to various sites
which were not active. That effort was only partially rewarded. Just one
failure, not the five needed to shut down the application pool for the site.

Event Type: Warning
Event Source: W3SVC
Event Category: None
Event ID: 1009
Date: 6/20/2005
Time: 9:07:11 PM
User: N/A
Computer: SELMA
Description:
A process serving application pool 'R0502523' terminated unexpectedly. The
process id was '1648'. The process exit code was '0xffffffff'.


That sounded from the help file like it should have triggered to tool to do
something. But there are no dump files -- just a dozen txt files and I
didn't see PID 1648 in any of them.

Should that have worked (produced a dump file)? Or must I catch it when it
kills an application pool completely?

Oh, by the way, I never saw that error in the browser. Everything came back
with the expected site or other normal (error) response. (Directory listing
denied for one case.)

mmr

"Bernard Cheah [MVP]" <qbernard@hotmail.com.discuss> wrote in message
news:uUfc1JMdFHA.1448@TK2MSFTNGP14.phx.gbl...
> Try debugdiag and hopefully it will able to find the root cause.
>
> --
> Regards,
> Bernard Cheah
> http://www.microsoft.com/iis/
> http://www.iiswebcastseries.com/
> http://www.msmvps.com/bernard/
>
>
> "M. M. Rafferty" <mmr@vistagrande.com> wrote in message
> news:O9xBWMJdFHA.1356@TK2MSFTNGP10.phx.gbl...
> > Hi Bernard,
> >
> > I would tend to agree that reading the metabase should not crash the app
> > pool. However, I have one instance where it happened to a site which is
> > temporarily set to redirect elsewhere -- nothing in the way of an IIS
app
> > to
> > debug as they still haven't uploaded the new version.
> >
> > None of these sites have any issues at any other time. Only when they
> > have
> > been shut down and the admin script has just executed does this happen.
> > There is nothing in the logs to provide a clue as to the specific
request.
> > Note that most of these are very small sites and have little use of ASP
or
> > ASP.NET except to supply the common header and footer elements.
> >
> > It must be what the admin script is doing, or something else with IIS6
or
> > Win2K3 is broken.
> >
> >
> >
> > mmr
> >
> >
> > "Bernard Cheah [MVP]" <qbernard@hotmail.com.discuss> wrote in message
> > news:e$ksfIIdFHA.3280@TK2MSFTNGP09.phx.gbl...
> >> Not sure, but i don't think reading metabase will cause app pool to
> >> crash.
> > I
> >> would suggest you get DebugDiag and try to debug the app.
> >>
> >> --
> >> Regards,
> >> Bernard Cheah
> >> http://www.microsoft.com/iis/
> >> http://www.iiswebcastseries.com/
> >> http://www.msmvps.com/bernard/
> >>
> >>
> >> "M. M. Rafferty" <mmr@vistagrande.com> wrote in message
> >> news:uoauLYHdFHA.1448@TK2MSFTNGP14.phx.gbl...
> >> > We are having an issue with the application pool being disabled due
to
> >> > rapid-fail protection. My best guess is that a routine admin script
> > which
> >> > is supposed to walk the metabase and copy the previous day's logs to
an
> >> > appropriately named folder on an archive box for stats, is locking
the
> >> > metabase. So far, I have been unsuccessful at correcting the
problem.
> >> >
> >> > Basic situation is a stand alone shared hosting box set with each
> >> > customer's sites being set up in an account specific application
pool.
> >> > (Might be more than one virtual server per application pool depending
> >> > on
> >> > the
> >> > customer.) The application pools are set to shutdown worker
processes
> >> > after
> >> > being idle 20 minutes. Some of the sites are just not very busy so
> >> > they
> >> > do
> >> > shut down.
> >> >
> >> > And it is those which seem to be having the application pool failure.
> > The
> >> > W3svc log for every one that has failed shows no activity for an
> > extended
> >> > time. The System log shows Event ID 1009 five times in less than a
> >> > second,
> >> > ("A process serving application pool 'xxxxxxxx' terminated
> >> > unexpectedly.
> >> > The
> >> > process id was '1604'. The process exit code was '0xffffffff'.") and
is
> >> > then
> >> > followed by event 1002 shutting down the application pool. It looks
> > like
> >> > a
> >> > spider, script kiddie or random visitor timed their request perfectly
> >> > so
> >> > the
> >> > scheduled task was finishing.
> >> >
> >> > Last night, the Application pool failures were 9:08:26PM to
9:08:27PM.
> >> > The
> >> > scheduled task running the admin script started at 9:07:00 PM and was
> >> > logged
> >> > off at 9:08:30PM. Actually, it should have been done by 9:07:09PM --
> > that
> >> > is when UPHClean logged an entry it had done cleanup. The script is
> >> > using
> >> > ADSI, nothing really strange, nothing that should be changing the
> >> > metabase.
> >> > But from the evidence of the crashing application pool, seems to be
> >> > locking
> >> > the metabase anyway.
> >> >
> >> > [section of code where it starts finding the sites -- just in case I
am
> >> > not
> >> > explaining this script adequately]
> >> > Set objW3SVC = GetObject("IIS://LOCALHOST/W3SVC")
> >> > 'Walk through the metabase, and pick out web servers only.
> >> > For Each objSITE in objW3SVC
> >> > If objSITE.class = "IIsWebServer" Then
> >> > servNum = objSITE.name
> >> > For each port in objSITE.ServerBindings
> >> >
> >> >
> >> > I could not see anything in the documentation indicating that this
> > should
> >> > be
> >> > a concern. Can someone help and point me to relevant docs? Is there
a
> >> > read-only way to do this. Or is this a bug with ADSI and IIS 6? Are
> >> > there
> >> > alternatives that would avoid the locking?
> >> >
> >> >
> >> >
> >> > Thanks.
> >> >
> >> > Mary Rafferty
> >> >
> >> >
> >>
> >>
> >
> >
>
>