Forums

Home / Forums

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

Node freezes when adding to cluster

Pages: 1 2

we have a node that when an attempt it made to add it to a cluster it freezes.  no kernel panic just a straight up freeze.  the config is 2 45 drives stornators and 1 other intel box.

things we have tried.

  1. creating the cluster with each of the different machines first and adding in different orders
  2. the 3rd node does the same thing virtual or physical
  3. installing petasan on different physical disks
  4. creating an all virtual 3 node cluster (works no problem)
  5. adding the nodes together with only management and nothing else
  6. adjusting the IP ranges in case i missed an IP conflict
  7. any combination of the above

the only thing we can get to work is 3 virtual nodes #4

there are no errors it just stops

has anyone elses seen this or have any advice?  i have 20 - 18 tb just waiting to get used and filled

 

thanks

Can you try to build the cluster without specifying any OSDs or any default pools. This eliminates a lot of startup functions. Once the cluster builds you can then via the management ui add the OSDs manually and pools, it will be easier to detect any freezing and on which function.

i know you checked your network configuration, ips, subnets ( make sure the subnets do not overlap in their ranges) but typically these are the most often issues.

thanks...did the double check on IP ranges all good.  even verified all the MTU on switch all matched

also I had tried building the cluster with only the management role on all 3 nodes and the issue still persists.  currently 2 of the nodes are in the cluster in management role only waiting for the 3rd management node.  but sadly it still locks the 3rd node.

any other thoughts we could try?

can you email the file /opt/petasan/log/PetaSAN.log on all 3 nodes to contact-us @ petasan.org

please send the logs of all 3 nodes after an unsuccessful attempt building the cluster in node 3 final step, wait 15 min then grab the logs.

also please include the file

/opt/petasan/config/cluster_info.json

Quote from admin on April 16, 2021, 8:27 pm

can you email the file /opt/petasan/log/PetaSAN.log on all 3 nodes to contact-us @ petasan.org

please send the logs of all 3 nodes after an unsuccessful attempt building the cluster in node 3 final step, wait 15 min then grab the logs.

also please include the file

/opt/petasan/config/cluster_info.json

sent, thanks!

from the logs in looks like there were prev, attempts to install. can you try once more installing from the beginning on all 3 nodes, using the installer to install os then deploy.

then if you it gets stuck in build of node 3, wait 15 min then send me same files + add the following for node 3 only:

dmesg

/var/log/syslog

just sent with the new logs and the syslog

did a clean install of all 3 nodes, management only, still #3 froze, waited 20 mins and sent logs

thakns

shawn

it is most probably a hardware issue with 3rd node, i would recommend you try with different host if you can, or to verify this switch the node order for it to become the second node and most probably the hang will occur there.

from syslog, it appears to freeze during sync of the RTC real time clock

hwclock --systohc --utc

the syslog also show a reboot after this freeze but not sure if you did that or the system did it itself.

Hello, i did the reboot to get the logs.  was not able to access it to grab logs after the attempt to add it to the cluster.

I have tried with 3 different boxes as the 3rd node-2 physcial and 1 virtual and the same thing the 3rd node locks up.   when we did 3 all virtual node we could create the cluster, which i thought could have been an issue with the 2 other nodes.

 

can you run the above command via ssh and see if it is causing the lock

Pages: 1 2