Jump to content
  • 0

corrupted json files


Andrew Gresbach1709152664

Question

We ran into a pretty big issue earlier today (and possible same thing a couple of weeks ago) where we had all of our MCS desktops which use full user layer (and other elastic layers) crash on us.  It took a while before we started to be able to launch sessions again and for a while after that we'd get in but get app layering errors like below:

image.thumb.png.1d1daf373ee8ebf6b0585039dd267fb8.png

 

I went into our smb share and restored previous version from the day before and would try again to get the same error but pointing to a different .json file in that share. eventually i restored every one of them and after a bit longer i was able to get in fine again.  It took quite a while for everyone to be able to get back in after that (probably login storm i'm guessing) but theres some big concerns what happened (and possibly happened a couple of weeks ago as well.....in that past case i didnt do anything and it came back on its own but similar symptoms).  We did not see anything that jumped out on the smb share side of things so no idea what to do w/ this but big concerns it could happen again and bring the system down again.  I opened a ticket and uploaded ELM logs but wanted to check in here as well to see if anyone has any ideas for us

 

thanks for the help!

Link to comment

5 answers to this question

Recommended Posts

  • 0

no unfortunately in the heat of it all we were just trying to do whatever we could to get things back up and running since it brought the whole company down.  i'll remember to do that next time.   and nope just the one elm in our environment

 

it has everyone a bit nervous though since this type of thing happened twice in the past month though and no idea why

Link to comment
  • 0

I was thinking a little more about this and the crashes make me think you might be having a storage or network issue.  The only time we read those files is during logon.  Once the user layer is mapped the files are no longer used.  So if you have machines crashing, I would think it must mean the storage is disconnecting for some reason.

Link to comment
  • 0

interesting........ I had a similar thought on that (that the json's SHOULDN'T get referenced once a session is loaded and i was confused why those would cause an issue w/ sessions .   why they were showing as corrupted makes me a little leary though but is there any harm in renaming them all and letting elm regenerate just to make sure they are fresh/new? i'm sure that may have nothing to do w/ it but just for good measure?

 

i sent a message to one of my open Nutanix tickets to see what they thought on it too

 

As far as the smb share like we've talked we currently are using Nutanix Files (currently only using a general share setup vs distributed due to limitations on the app layering requirements of needing that single root "Users" folder).  but looking how in the meantime we can get this more redundant or stable.....its been suggested to look at DFS so it can distribute the load evenly across the 3 file servers we have but our current DFS setup is already using that "users" namespace so that kind of kills that idea i think.   another forum member mentioned using DFSR (active / active) and that should help w/ user layer redundancy but i think even w/ that setup if one of the nodes would blip or go down user layers still would drop right? (given how critical to the system they are)

 

Link to comment
  • 0

The Elm should be recreating those files I think its every 200 minutes.  If you change permissions it will update them as well.  Next time it happens I think you need to get a copy before fixing it so that we can try to see what corrupted means.  You can also check with support to see what they think.

 

As for DFSR it wouldn't probably help here and in general its not helpful for availability with the user layer because there is no automatic failover of existing connections like you have with a cluster.  Plus its not like your file server failed here it was still working. 

 

For the user layer you can set up as many shares as you want and split the users between shares by AD Group membership.  That's done in the System Tab under User Layer Storage Locations.  You can also use the registry to define alternate elastic layer shares for the json files and split that up.  You can find out info on that here https://www.citrix.com/products/citrix-virtual-apps-and-desktops/resources/understanding-elastic-layering.html#tab-37495 under scaling.

 

HKEY_LOCAL_MACHINE\SOFTWARE\Unidesk\ULayer\RepositoryPath

Value = \\unideskfs1\unidesk

 

Rob

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...