Website VU feils - looking for ideas

[aldets]

Hi all,

I'm just looking for ideas where I could look further.

The thing is that for some reason all the Website VU's for every installation won't come up on a secondary node. Everything works just peachy on node 1, but whenever I move the Website VU to node number 2, it just feils. Every other VU moves just fine between nodes. It looks like it isn't even trying much, because the VU doesn'y generate any logs (exept for EmailSender, but that's not feiling).

I've reinstalled the prerequisite software. The server itself has switched Host, but nothing. I think there might be something on the IP level, but can't be sure. IP comes up for a brief moment on node 2 and then it goes back to node 1 and leaves a failed VU behind.

Tryed uninstall/install the VU, nothing.

Luckily this is a test environment. Unluckily I have no idea when this all started as we don't move VU's in the test environment practically at all.

So if you have any tips or tricks where I could look, it would be very great.

Kind regards,

Alder Tsvetkov

Application administrator

[dhlashwe]

Hi Aldets,

this is really funny, we have EXACTLY the same cases. Two are with customers, where we didn't do the original installation, so we did not have the environment fully under control.

But just recently, we were able to reproduce the issue in our neat and clean lab environment. Just adding new instance works for all other VUs but website/webclient based...

We have open ticket for this already, so I suggest you to do the same.

BR,

Dawood

[aldets]

Ok, that truly sounds odd. Could you share what version are you running? Our installations are FP17 and FP18 (hotfixes applied).

I can remove and add instance without a problem. Once I move the Website VU to second instance, it fails. I think the problematic software is WebClient. We have one installation with minimum amount of software and under HAC it can be observed that WebClient fails, but Email Sender is atleast trying, since it produces some lines to the logs.

Alder

[Jukka]

Sounds very odd and brings in mind GARP driver we used to had some years back. Could either of you create a ticket to get this into official investigations? If possible it doesn't make harm to try out with GARP driver. 

//Jukka

[aldets]

Hi Jukka,

I've made a ticket last week - SCC-3482.

The GARP is a bit confusing for me and raises some questions. I think I haven't used it in ages. And it's strange that it should be used now. The issue just appeared out of the blue. I can't say for sure when it all happened.

Alder

[dhlashwe · Accepted solution]

Hi Aldets, Jukka,

my colleague has fixed the problem, he had to go really deep into the IIS tracing. We had the problem on two FP16P100 and one FP17 and there was no single cause. In one case, he changed only port to some other than 80 and then back and it works. Second case, there was missing some DLL (client-server), it was necessary to add via features. Third case was some mess in handlers mapping.

I think in such cases, reinstall of VU is not enough, it is better to get rid of IIS itself from the OS and then add it again.

It is funny that these cases appeared recently, yet they have different root causes...

Lasse from Sinch support gave us some very good hints on how to trace down and diagnose the problem.

Hope this helps a bit.  

Dawood

[aldets]

Hi,

Good news, one of your suggestion helped me out. I went straight for the IIS kill. Removed it as role from the virtual machine. Restarted it and then added it back. Did Upgrade all Software for the Website VU on secondary node and now everything is peachy.
Thanks for the info.

Alder