Connectivity Bug? 9.1R2 2331 - Anyone else?

[jpjackso]

Last week we updated our PSA5000 cluster to PCS 9.1R2

Ever since, we've started getting a small rate of user complaints about having NO working network access after signing onto the appliance with various versions of the PCS client.  We're still working to get a clear definition of who and what is involved.  So far, it's been mostly, if not all, Macintosh client users.  Not certain yet if it's just Macs though.  Also, sometimes the problem resolves on its own, after the user logs out for a few hours and tries again.  Sometime, if they don't wait that long and seek assistant from our internal support group, uninstalling and reinstalling the PCS client seems to resolve the problem, at that moment.  Not sure yet if it returns for such a user then later, in that case.

What does seem to be clearly correlated is that this wasn't happening when we were on 9.1R1 1505.

Is anyone else here seeing anything similar?  I see a post here from Monday that could match, but it's for an older version of PCS that we haven't used in months.

-John

[jpjackso]

This is definitely an obnoxious bug.

I just focused in on one user with the issue.  One a Windows 10 machine, Pulse client, that user logs in to the PCS5000 and then can not pass traffic to anywhere in the ip network the appliance is configured to give him access to.  As soon as he disconnects the session, he has working access to that network again.

If I login to the appliance, on the same machine, I have no problem accessing the mentioned network.  His broken access, versus it working for me, could not be corrected by anything, including removing the Pulse client completely from the machine, rebooting and reinstalling it.  Problem remained.  Clearing the user's session from the appliance side made no difference.

If the user connects to the appliance from another Windows machine, he has no trouble at all, on that one, or from a Macintosh, as well.  It's only when he connects to the appliance from his own machine, and it's only when he connects, not when someone else uses the same machine and their own credentials.

I'm inclined to believe there's an issue on the appliance side where it's got something cached that tied to this user, from this machine, based on something it picks up from the Pulse client, which persists even after completely wiping out the client install and reinstalling from scratch.

As far as I know, this issue wasn't cropping up before the 'upgrade' on the appliance from 9.1R1 to 9.1R2.  I'm strongly feeling like just rolling back to 9.1R1 and hoping the developers will find and fix whatever's going on by a later release.

Is roll back as simple as it appears, in the admin interface?

-John

[jpjackso · Accepted solution]

I insist on coming clean when I discover I broke something!! 

There was no bug.

There was an ip address conflict with another device on the network that the users were being assigned addresses out of, when connecting to the VPN appliance.  I figured this out after realizing that the users who were randomly experiencing the problem were all being assigned the same ip address when they did.  I eliminated the address conflict and now.... paradise.

[[email protected]]

Thank you for taking your time to share this