Forums

Home / Forums

You need to log in to create posts and topics. Login · Register

Petasan 2.1 Installation

Pages: 1 2

andreas
16 Posts
October 9, 2018, 7:41 am

Thank you for releasing version 2.1.

I'm having trouble installing 3 Management Nodes without OSDs and iSCSI. The last step of the 3rd node does not come to an end. If I abort the installation and restart all three nodes, only the third node will show that the installation is complete. I can still sign in to the Mangement Console and all three monitors seem to be running correctly.

Thanks Andreas

#1

admin
2,930 Posts
October 9, 2018, 11:36 am

This works in our testing. The third node deployment may wait for 5 min or less for any storage pools to come up, but it is a very small wait. I suggest you re-install from beginning and do not abort.

When done you will be able to log via the web ui, some  pages work well, some will feel a bit slow (about 5 sec response)  since many storage related operations time out in the background. Once you add storage, your default storage pool will be active and all storage related pages will be fast.

#2

andreas
16 Posts
October 9, 2018, 1:12 pm

I have tried again and waited about 10 minutes, unfortunately without success.
After that, I configured two OSDs at each management node. Even now, the last step was not completed. After the reboot ran the monitors and the OSDs. But no pool was created.

#3

admin
2,930 Posts
October 9, 2018, 1:38 pm
  1. So if understand correctly you did choose 1 osd in the first 2 nodes and still the cluster build failed..correct ?
  2. Does the build step  return an error or does it just hang forever ?
  3. Can all nodes ping each other on management, backend 1 and backend 2  networks ?
  4. What hardware are you using ? how much RAM ?
  5. If you are using low end hardware...how much time do you wait in the build step..can you allow 15 min before aborting ?
  6. Can you log in the web ui  ? what is the status ? if its not "ok", what does it say (click on it to get details) ? What are the number of OSDs : are they up or down ?
#4

andreas
16 Posts
October 9, 2018, 3:06 pm
  1. On each node one OSD. In sum 3 OSDs
  2. It hangs no error or time out
  3. Yes. After rebooting I can ping on all 3 networks each other. I have 3 network cards.
  4. virtual machines on vmware workstation. 2 CPUs, 3GB RAM, 32GB root and 16 GB OSD drive
  5. It runs on my laptop (4 CPUs, 32GB RAM)

After 10 min waiting I SSH to the first node an got the following ceph status.

 

Last edited on October 11, 2018, 5:46 am by andreas · #5

admin
2,930 Posts
October 9, 2018, 3:32 pm

My first suspect is the OSD disk size of 16GB is too low, try 32 or 64  GB. Also bump up the RAM to 4 GB. It is possible the OSD show they are up but due to their size operations like creating a pool would hang.

The min disk size per  our hardware  guide is 32 GB  and i believe we check for this in the installer but maybe we missed this validation in the deployment wizard.  i will log this as a bug to enforce this.

If this does not fix the issue,  do let me know and please post the log file /opt/petasan/log/PetaSAN.log for the third node after it hangs for 10 min.

#6

andreas
16 Posts
October 9, 2018, 5:35 pm

I resize the OSD disks to 50GB and bump the RAM to 4GB.

 

Last edited on October 11, 2018, 5:39 am by andreas · #7

admin
2,930 Posts
October 9, 2018, 5:59 pm

It is not clear for me: after you increased the disk size, do you still have node 3 hang in build + no pools created ? did you re-install via the installer iso after bumping the disks ?

note the errors in logs are not alarming, they relate to the stats server which takes time to come up.

Last edited on October 9, 2018, 6:01 pm by admin · #8

andreas
16 Posts
October 9, 2018, 6:24 pm

Oh, I added the OSD disks after the iso installer and before the installation wizard.

Tomorrow I try the whole installation with the resized OSD disks and come back.

Thanks for today

#9

andreas
16 Posts
October 10, 2018, 6:34 am

I tried a new fresh installation from iso.
3 nodes (2 CPUs, 4GB RAM, 32GB root disk, 50GB OSD disk) unfortunately with the same result as yesterday.

1. Monitors are generated and went up.
2. OSD are generated and started up
3. Monitor 2 is shut down again

Nothing happens anymore.

 

 

Last edited on October 11, 2018, 5:45 am by andreas · #10
Pages: 1 2