0

We have a production system where we are generating 1000s of XML files each having the max size of 150MB everyday to NFS storage mounted to our applications servers. Both NFS and application servers are running on Solaris 10 operating system.

We are using JAXB(Java XML API) to generate the XML files.

The application server have no errors reported but once a while generated XML files are corrupted like missing tags and not completed properly. We are suspecting the issue is because of the NFS storage as we have ruled out other possibilities of application issues(Not able to reproduce and rerunning the job generating the file without any issue).

Can anyone suggest how can we make sure there is no issue with NFS writing? or any other suggestion on how we can debug the issue?

Regards, Mayuran

Mayuran
  • 669
  • 2
  • 8
  • 39

1 Answers1

1

If you are consistently having NFS file corruptions, take a look at your network. It's likely that the network is unstable, thus causing writes to get lost in transit. I had that issue at a previous job, only to find out that the network was having so many collisions that some of our packets were getting thrown away due to timeouts, and NFS isn't very fault tolerant when it comes to network (or anything, for that matter).

First choice would be to not use NFS if possible, but aside from that, check your network with switch logs, tcpdumps, etc.

Tim S.
  • 2,187
  • 1
  • 25
  • 40