Upgrade from OSS 2.14.5 to 3.6.02 (process stops continously at "X minutes from last change")

We are upgrading Nexus from OSS 2.14.5 to 3.6.02 migrating 14 repositories using hardlink method.

This is our second second attempt, we had to cancel the first one because the inodes got full and we got a lots of fails. We canceles it at 78% after 4 days of execution.

We restarted the upgrade following the instructions in the doc:

  1. Shut down Nexus Repository Manager 2 and remove the entire NXRM 2 $work-dir /db/migrationagent directory.
  2. Shut down Nexus Repository Manager 3 and remove the entire NXRM 3 $data-dir directory.

Apparently this second attempt is being much slower than the first one, we have only 28% in 7 days.

We have observed the synchronization stucks frequently showing this message:

XXXXX processed, 0 minutes since last change, press “Continue” when ready

We understand that this message should only appear when all repositories had been synchronized, shouldnt it?

We also saw this messages in the first attempt, but in this one we see until 34 minutes of wait.

The only difference between both attemps is the config of Blobstores: in the first one we used only 2 blobstore for the 14 repos and in the second one we use one blobstore for each repo.

Any ideas why this second attempt is being so slow compared with the first one?

We’re glad to see you migrating to NXRM 3! Have you seen our Upgrading Guide? In a nutshell, I’d like to kindly ask you first to upgrade your NXRM2 to the latest version, and then migrate to the latest NXRM3 version. The versions you are currently using are both back from 2017 and we’ve made a number of improvements since!

Hi! we finally upgrade to latest versions:

Nexus 2 – OSS 2.14.18-01
Nexus 3 – OSS 3.23.0-03

The sinchronization is running.

We still see the synchronization stucks frequently showing this message (sometimes we see 4 minutes waiting in this state):

XXXXX processed, 0 minutes since last change, press “Continue” when ready

Is it normal?

Thanks Dawid!

We have upgraded to the latest versions. As you can read in my last post, the syncrhonization is still going too slow comparing to the first execution. A difference between our first attempt (it was fast, almost all the repos synchronized in 5 days) and the second and third attempts (both very slow) is the blobstores’ configuration: in the first attempt we created 2 blobstores for all the repositories (and blobtores’s path were out of the repositories path). In the second attempt we created one blobstore for each repository and the blobtores’ path are in the root folder of each repository. We dont know if this configuration could affect to the syncrho speed (second attempt was cancelled because of a full filesystem and maybe that truncated migration left some trash in the repos that are causing the current sloweness). However, as I said in the main post, when we have restarted the migration we have followed the instructions:

  1. Shut down Nexus Repository Manager 2 and remove the entire NXRM 2 $work-dir /db/migrationagent directory.
  2. Shut down Nexus Repository Manager 3 and remove the entire NXRM 3 $data-dir directory.

Any ideas what could be causing the sloweness?
Thanks in advance Dawid!