I've been using smpatch to patch my Solaris 10 systems. After having
used it for a while I've observed some odd behavior with it that make me
question if there is a problem with it:

1. When downloading patches with the "update" action the download
occasionally freezes. A control-c and a re-run of the update action
results in the remaining patches being downloaded and installed.

2. Occasionally after performing the update action and rebooting an
immediate run of smpatch with the "analyze" action shows that there are
some additional patches to be installed.

3. When using smpatch in a command window in JDS the console
occasionally displays java related errors.

While I have no exhibited any problems with the system after any of
these issues I have to wonder how reliable smpatch is. Is anyone else
experiencing these issues?

Josh

Josh McKee wrote:
> I've been using smpatch to patch my Solaris 10 systems. After having
> used it for a while I've observed some odd behavior with it that make me
> question if there is a problem with it:
>
>[...]

Indeed, there is a problem. I have three UltraSparc boxes with Solaris
10 installed and I am experiencing the same problem that you reported.
This almost always fails:

# smpatch update

I found out that doing an

#smpatch analyze"

and than

# smpatch update -i <patchId>

where patchId is the Id given in the first column of the patch list
given by "analyze", has more chances of success (and also allows to get
only the necessary patches).

Indeed, Solaris' "smpatch" is way behind Linux's "yum" when it comes to
applying patches. Also being smpatch java based put a small box like an
Ultra 5 on its nees, getting 100% of the CPU dor two or three minutes
at startup.

Given the chronic lack of absolute speed of UltraSparc processors, Sun
should make a lesser use of CPU hungry Java based utilities.

HD

Josh McKee wrote:
> I've been using smpatch to patch my Solaris 10 systems. After having
> used it for a while I've observed some odd behavior with it that make me
> question if there is a problem with it:
>
> 1. When downloading patches with the "update" action the download
> occasionally freezes. A control-c and a re-run of the update action
> results in the remaining patches being downloaded and installed.
>
> 2. Occasionally after performing the update action and rebooting an
> immediate run of smpatch with the "analyze" action shows that there are
> some additional patches to be installed.
>
> 3. When using smpatch in a command window in JDS the console
> occasionally displays java related errors.
>
> While I have no exhibited any problems with the system after any of
> these issues I have to wonder how reliable smpatch is. Is anyone else
> experiencing these issues?
>
> Josh

Give PCA a try:

http://www.par.univie.ac.at/solaris/pca

I've been thrilled with it, except for all the problems with
Sun's patch quality control and server problems (but those are
Sun issues). PCA is awesome.


Dave

In article <9odLg.1471$I71.58@trnddc01> Dave Littell <littelld_nospam@verizon.net> writes:
>Josh McKee wrote:
>> I've been using smpatch to patch my Solaris 10 systems. After having
>> used it for a while I've observed some odd behavior with it that make me
>> question if there is a problem with it:
>>
>> 1. When downloading patches with the "update" action the download
>> occasionally freezes. A control-c and a re-run of the update action
>> results in the remaining patches being downloaded and installed.
>>
>> 2. Occasionally after performing the update action and rebooting an
>> immediate run of smpatch with the "analyze" action shows that there are
>> some additional patches to be installed.
>>
>> 3. When using smpatch in a command window in JDS the console
>> occasionally displays java related errors.
>>
>> While I have no exhibited any problems with the system after any of
>> these issues I have to wonder how reliable smpatch is. Is anyone else
>> experiencing these issues?
>>
>> Josh
>
>Give PCA a try:
>
>http://www.par.univie.ac.at/solaris/pca
>
>I've been thrilled with it, except for all the problems with
>Sun's patch quality control and server problems (but those are
>Sun issues). PCA is awesome.
>
>
>Dave

Smpatch on my old Ultra 1E (170 MHz), running Solaris 8 takes 5 minutes
to run. Pca takes less than a minute to run its patch analysis, and it
just needs Perl (instead of Java).
--
Jeff Wieland

What are the java error messages?

Do you have enough disk space?

When does you job starts? Maybe you should try a different piont in
time.

Are there some network problems.


Josh McKee schrieb:

> I've been using smpatch to patch my Solaris 10 systems. After having
> used it for a while I've observed some odd behavior with it that make me
> question if there is a problem with it:
>
> 1. When downloading patches with the "update" action the download
> occasionally freezes. A control-c and a re-run of the update action
> results in the remaining patches being downloaded and installed.
>
> 2. Occasionally after performing the update action and rebooting an
> immediate run of smpatch with the "analyze" action shows that there are
> some additional patches to be installed.
>
> 3. When using smpatch in a command window in JDS the console
> occasionally displays java related errors.
>
> While I have no exhibited any problems with the system after any of
> these issues I have to wonder how reliable smpatch is. Is anyone else
> experiencing these issues?
>
> Josh

Denis wrote:
> What are the java error messages?
>
> Do you have enough disk space?
>
> When does you job starts? Maybe you should try a different piont in
> time.
>
> Are there some network problems.
>
>
> Josh McKee schrieb:
>
> > I've been using smpatch to patch my Solaris 10 systems. After having
> > used it for a while I've observed some odd behavior with it that make me
> > question if there is a problem with it:
> >
> > 1. When downloading patches with the "update" action the download
> > occasionally freezes. A control-c and a re-run of the update action
> > results in the remaining patches being downloaded and installed.
> >
> > 2. Occasionally after performing the update action and rebooting an
> > immediate run of smpatch with the "analyze" action shows that there are
> > some additional patches to be installed.
> >
> > 3. When using smpatch in a command window in JDS the console
> > occasionally displays java related errors.
> >
> > While I have no exhibited any problems with the system after any of
> > these issues I have to wonder how reliable smpatch is. Is anyone else
> > experiencing these issues?
> >
> > Josh

On Mon, 04 Sep 2006 08:46:25 -0600, Josh McKee wrote:

> I've been using smpatch to patch my Solaris 10 systems. After having
> used it for a while I've observed some odd behavior with it that make me
> question if there is a problem with it:
>
> 1. When downloading patches with the "update" action the download
> occasionally freezes. A control-c and a re-run of the update action
> results in the remaining patches being downloaded and installed.

That's Sun's servers crashing.
I've watched the network traffic. Their server suddenly goes down.

Another thing I don't like is after a host is running for several
weeks I'll visit it and find several smpatch processes running but
idle. I guess Sun's servers have crashed during a cron job and these
processes are left. I find killing off these things and their process
trees makes a new smpatch run work much better.

>
> 2. Occasionally after performing the update action and rebooting an
> immediate run of smpatch with the "analyze" action shows that there are
> some additional patches to be installed.

Yes, notice that too. Also notice some times a patch says it is
required yet when it is installed it says the software package
the patch is made for is not installed on this host.

> While I have no exhibited any problems with the system after any of
> these issues I have to wonder how reliable smpatch is. Is anyone else
> experiencing these issues?

Seems patching is getting flakier as time goes on. Faster, better,
cheaper. Choose any two.

Regards, Scott

In article <1157523957.962559.180680@d34g2000cwd.googlegroups .com>,
"Denis" <denis.nicklas@anic.info> wrote:

> What are the java error messages?

I don't know as I didn't take note of the specific errors. Just noted
that they occurred.

> Do you have enough disk space?

Plenty.

> When does you job starts? Maybe you should try a different piont in
> time.

It's run manually.

> Are there some network problems.

Doubtful as it's happened on multiple systems: at work, home, elsewhere.

Josh

> Josh McKee schrieb:
>
> > I've been using smpatch to patch my Solaris 10 systems. After having
> > used it for a while I've observed some odd behavior with it that make me
> > question if there is a problem with it:
> >
> > 1. When downloading patches with the "update" action the download
> > occasionally freezes. A control-c and a re-run of the update action
> > results in the remaining patches being downloaded and installed.
> >
> > 2. Occasionally after performing the update action and rebooting an
> > immediate run of smpatch with the "analyze" action shows that there are
> > some additional patches to be installed.
> >
> > 3. When using smpatch in a command window in JDS the console
> > occasionally displays java related errors.
> >
> > While I have no exhibited any problems with the system after any of
> > these issues I have to wonder how reliable smpatch is. Is anyone else
> > experiencing these issues?
> >
> > Josh

I noticed to that Sun's patchserver had availability problems but the
last 2 months I had no trouble anymore.

Scott Packard schrieb:

> On Mon, 04 Sep 2006 08:46:25 -0600, Josh McKee wrote:
>
> > I've been using smpatch to patch my Solaris 10 systems. After having
> > used it for a while I've observed some odd behavior with it that make me
> > question if there is a problem with it:
> >
> > 1. When downloading patches with the "update" action the download
> > occasionally freezes. A control-c and a re-run of the update action
> > results in the remaining patches being downloaded and installed.
>
> That's Sun's servers crashing.
> I've watched the network traffic. Their server suddenly goes down.
>
> Another thing I don't like is after a host is running for several
> weeks I'll visit it and find several smpatch processes running but
> idle. I guess Sun's servers have crashed during a cron job and these
> processes are left. I find killing off these things and their process
> trees makes a new smpatch run work much better.
>
> >
> > 2. Occasionally after performing the update action and rebooting an
> > immediate run of smpatch with the "analyze" action shows that there are
> > some additional patches to be installed.
>
> Yes, notice that too. Also notice some times a patch says it is
> required yet when it is installed it says the software package
> the patch is made for is not installed on this host.
>
> > While I have no exhibited any problems with the system after any of
> > these issues I have to wonder how reliable smpatch is. Is anyone else
> > experiencing these issues?
>
> Seems patching is getting flakier as time goes on. Faster, better,
> cheaper. Choose any two.
>
> Regards, Scott

Scott Packard wrote:
[deletia]
>
>
> Seems patching is getting flakier as time goes on. Faster, better,
> cheaper. Choose any two.
>

Really? Let's work it through:

Faster? Based on what I've seen, you have about a 30% chance
that any given patch will fail to download. So you get to try it
again, only to find:

Better? Patch quality is deteriorating almost daily and each S10
update is less likely to (really) work out-of-the-box. Your odds
of bricking your system are excellent if you blindly patch.

Cheaper? With a pay-for-support revenue model, you pay for
access to all those patches that just might brick your system.

Conclusion: patch only when desperate.


Dave