[olug] Automating killing a *nix process from a windows box
troehr at nj-onramp.com
troehr at nj-onramp.com
Wed Mar 24 01:23:38 UTC 2010
The one box solution is desirable, but you also are now looking at
modifying what was probably a vendor provided configuration...
That tends to open up a new can of worms like anything that breaks
after that on the Solaris/GoldenGate systems will be blamed on your
besides... how hard is
tail -f logfile|grep locked process pid|xargs -n1 ssh
user at solarissystem kill -9
*of course, ther might be a little more then that... but it shouldn't
be much more*
Quoting "T. J. Brumfield" <enderandrew at gmail.com>:
> I completely agree, but we are such a Microsoft shop, that the few
> people that are responsible for *nix systems seem to resent it, and
> barely seem to know those systems.
> Unfortunately, I don't own those boxes so I'm somewhat at their mercy.
> I'm just responsible for supporting an application which depends on
> databases on those boxes.
> -- T. J.
> On Tue, Mar 23, 2010 at 11:04 AM, Christopher Cashell
> <topher-olug at zyp.org> wrote:
>> On Tue, Mar 23, 2010 at 10:41 AM, T. J. Brumfield
>> <enderandrew at gmail.com> wrote:
>>> Here is the problem. We have error logs spitting out on Windows boxes
>>> about GoldenGate processes locking up on a Solaris box. This is a
>>> known bug that GoldenGate is hoping to fix in their next release. The
>>> Solaris team has told me these logs aren't visible from the Solaris
>>> boxes, despite the fact that is where GoldenGate is actually running.
>>> I don't exactly believe that, but I have to go off what they tell me.
>> How do the logs or log data get from the Solaris box to the Windows
>> box? Is it directly from a GoldenGate process on the Solaris box to
>> some sort of GoldenGate process on the Windows box (not familiar with
>> GoldenGate, so no idea how it operates or what it does)? If it is
>> purely within the application, then it is possible that the logs are
>> never hitting the local system (maybe GoldenGate could be configured
>> to log locally?). However, if the logs are being transported via some
>> non-application method (like syslog), then it should be possible to
>> get the logs to the local system, even if they aren't being directed
>> there now.
>> Alternately, is there any way at all to recognize the locked process
>> from the local Solaris box? Something other than the logs on the
>> Windows box?
>> If it were me, I'd exhaust all possible solutions that could be
>> contained on the Solaris box, just because it "feels" like solving it
>> there would be a lot cleaner and more reliable (fewer moving parts).
>>> -- T. J. Brumfield
>> OLUG mailing list
>> OLUG at olug.org
> OLUG mailing list
> OLUG at olug.org
More information about the OLUG