Notification if Indexing or Publishing is down?

[NATHANBROWN]

We have PDMLink 9.1 and are having issues with publishing and indexing going down. We have a separate workstation that is used to publish our CADD files andthe workstationhas Fast InStream for indexing. The issue is that we do not know when the publisher or indexer go down. Our main concern is the indexer. If files are not being indexed, then older rev files will be found from a global search. Then someone could check out an older rev file and make changes, not knowing there is a newer rev out there. We have stopped using our global search for now. It would be nice to get a notification if the publisher or indexer go down. Otherwise, it could be a week or two later and then we have to reindex the whole vault.

[MikeLockwood]

This doesn't solve the problem with indexing going down, but FYI we developed a query builder report that pulls the index status for each iteration of a User-entered document Number - to confirm whether any iterations have in fact not been indexed successfully. I'll send the qml (9.0 M050) if you like.

[jfrankovich]

Nathan,

Have you found a solution to your problem yet? We are looking for a way to just monitor our Publisher and get notification of failed jobs by email back to an admin. Our publisher is pretty stable, but on the occasion when it does fail a job, it usually fails everything from there on and we don't find out until the next morning.

John Frankovich

The GSI Group LLC

In Reply to Nathan Brown:

We have PDMLink 9.1 and are having issues with publishing and indexing going down. We have a separate workstation that is used to publish our CADD files andthe workstationhas Fast InStream for indexing. The issue is that we do not know when the publisher or indexer go down. Our main concern is the indexer. If files are not being indexed, then older rev files will be found from a global search. Then someone could check out an older rev file and make changes, not knowing there is a newer rev out there. We have stopped using our global search for now. It would be nice to get a notification if the publisher or indexer go down. Otherwise, it could be a week or two later and then we have to reindex the whole vault.

[NATHANBROWN]

Yes, we did find a way to get notifications if publishing jobs fail or the publisher goes down. It is called the JMX Email Function. TAN138107 explains how to get it set up. This is exactly what we were looking for. JMX also sends me notifications if there are any errors with the Windchill server (ex. memory issues or the MethodServer goes down). Unfortunately, not many people know about JMX. I am having an issue adding multiple people to the JMX email list. I can only get one to work. PTC tech support is looking into it, but we have tried every syntax we can think of and it still only accepts the first email address. The email addresses should be separated by a comma according to the Windchill system admin guide, but TAN138107 shows a period separating the email addresses. There also is a way to go into the Java Console and add email addresses to JMX notifications, but the changes are replaced with the original value the next day. So just put in the one email address and it should work fine.

[sdrzewiczewski]

After you configure JMX Email per the TAN you listed, where do you
configure what to be notified of?

Thanks,
Steve D.

> Yes, we did find a way to get notifications if publishing jobs fail or the
> publisher goes down. It is called the JMX Email Function. TAN138107
> explains how to get it set up. This is exactly what we were looking for.
> JMX also sends me notifications if there are any errors with the Windchill
> server (ex. memory issues or the MethodServer goes down). Unfortunately,
> not many people know about JMX. I am having an issue adding multiple
> people to the JMX email list. I can only get one to work. PTC tech
> support is looking into it, but we have tried every syntax we can think of
> and it still only accepts the first email address. The email addresses
> should be separated by a comma according to the Windchill system admin
> guide, but TAN138107 shows a period separating the email addresses. There
> also is a way to go into the Java Console and add email addresses to JMX
> notifications, but the changes are replaced with the original value the
> next day. So just put in the one email addr
> ess and it should work fine.
>
>

[jessh]

Any number of e-mail addresses will work.

If using the Ant script, you just need to figure out the appropriate
quoting syntax for your shell to treat the -Dxxxx=yyy,... argument as
one argument.

Also if you edit the e-mail address list via jconsole (aka Java Console)
or VisualVM, then you can select the Loader's save() operation to save
your changes. The reason the script exists is to do this modification
to the method server, server manager, and servlet engine JMX
configuration all via one command rather than having to do the setting
in 3 places yourself.

--
Jess Holle
>
> Yes, we did find a way to get notifications if publishing jobs fail or
> the publisher goes down. It is called the JMX Email Function.
> TAN138107 explains how to get it set up. This is exactly what we were
> looking for. JMX also sends me notifications if there are any errors
> with the Windchill server (ex. memory issues or the MethodServer goes
> down). Unfortunately, not many people know about JMX. I am having an
> issue adding multiple people to the JMX email list. I can only get
> one to work. PTC tech support is looking into it, but we have tried
> every syntax we can think of and it still only accepts the first email
> address. The email addresses should be separated by a comma according
> to the Windchill system admin guide, but TAN138107 shows a period
> separating the email addresses. There also is a way to go into the
> Java Console and add email addresses to JMX notifications, but the
> changes are replaced with the original value the next day. So just
> put in the one email address and it should work fine.
>
>

[jessh]

P.S. To be clear the separator should be a comma, not a period.

[jessh]

There are a number of NotificationHandler MBeans configured
out-of-the-box to e-mail to the JMX-Administrators e-mail list in the
method server, server manager, and servlet engine JVMs.

You can modify the configuration of these MBeans or add additional
NotificationHandler MBeans. In any case a NotificationHandler MBean
listens for a JMX-level notification (event), so an MBean must exist
which generates the JMX notification of interest in any case.
NotificationHandlers are already provided for what were seen as the most
important JMX notifications.

One can also configure additional "user-specified" monitor MBeans via
the UserSpecified MBean in the Monitors section of the MBean tree.
These MBeans poll other MBean's attributes and fire a JMX notification
when some criteria is met, e.g. when a numeric attribute is too high or
too low.

--
Jess Holle

[jsteinbach]

Since we are on the subject of notifications, is there a way to monitor
the status of vaults? My vault goes to "read only" every once in a
while, not sure why just yet. If I could be notified when it happens I
might be able to determine the cause.

Does anyone know how to monitor the read/write status of a vault?

Thanks,
John