2

My company literally has thousands of NFS volumes, which we need to mount on a few of our servers. However, we never mount all of them at the same time; typically we mount about 12 at a time. Most of the time, most of the NFS volumes are actually offline. In the past, we have been mounting and unmounting them manually via the command line, as the need arose.

We would like to move to using autofs for mounting these volumes.

Is it bad to define thousands of autofs rules, one for each NFS volume? Does autofs have a hard limit about the number of rules it can have? Does autofs performance decrease drastically with a large number of rules?

1

We have several maps, and some of the maps have 10s, 100s of volumes. So it does work for a fleet of 100s of Linux, 10s of other platforms. The only issue we had with old autofs software in the early days was that it could not cope with modifications and updates online. That is changing the NFS volume and mount point during decommissioning old NFS filers. I do not believe performance is an issue. However, one always has to test and confirm. Scalability and performance were enhanced with Linux Autofs 5.x after introducing multithreading. For us our minimal distro is RHEL6 comes already with 5.x. so no it is not bad, you need to benchmark and test it, it will add better flexibility and stability to your environment as the configuration will be centralised "I am hoping that is one reason you would like to do so."

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.