You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ozone.apache.org by "Sadanand Shenoy (Jira)" <ji...@apache.org> on 2024/06/06 10:24:00 UTC
[jira] [Updated] (HDDS-10984) Tool to restore SCM certificates from RocksDB.
[ https://issues.apache.org/jira/browse/HDDS-10984?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Sadanand Shenoy updated HDDS-10984:
-----------------------------------
Description:
Although not a common scenario, but recently encountered an issue where the user accidentally deleted the SCM certs from the local certs dir from All SCMs.
On restart, the primordial node self signs a new cert and starts up however the non-primordial members in the HA ring, request a CSR to the primordial node. This request doesn't go through as it requires a quorum i.e ratis needs to be initialized . however in order for ratis to be initialised, it needs to verify certs. This leads to a dead-lock like situation.
Ratis is needed for CSR (getCert) request as it persists the signed Cub ca cert in its rocksdb.
Like I mentioned in the beginning, this is not a common case but since we already persist certs in the SCM rocksdb, a tool to restore certs should be helpful in this case.
was:
Although not a common scenario, but recently encountered an issue where the user accidentally deleted the SCM certs from the local certs dir from All SCMs.
On restart, the primordial node self signs a new cert and starts up however the non-primordial members in the HA ring, request a CSR to the primordial node. This request doesn't go through as it requires a quorum i.e ratis needs to be initialized however in order for ratis to be initialised, it needs to verify certs. This leads to a dead-lock like situation.
Ratis is needed for CSR (getCert) request as it persists the signed Cub ca cert in its rocksdb.
Like I mentioned in the beginning, this is not a common case but since we already persist certs in the SCM rocksdb, a tool to restore certs should be helpful in this case.
> Tool to restore SCM certificates from RocksDB.
> ----------------------------------------------
>
> Key: HDDS-10984
> URL: https://issues.apache.org/jira/browse/HDDS-10984
> Project: Apache Ozone
> Issue Type: Bug
> Components: SCM HA, Tools
> Reporter: Sadanand Shenoy
> Assignee: Sadanand Shenoy
> Priority: Major
>
> Although not a common scenario, but recently encountered an issue where the user accidentally deleted the SCM certs from the local certs dir from All SCMs.
> On restart, the primordial node self signs a new cert and starts up however the non-primordial members in the HA ring, request a CSR to the primordial node. This request doesn't go through as it requires a quorum i.e ratis needs to be initialized . however in order for ratis to be initialised, it needs to verify certs. This leads to a dead-lock like situation.
> Ratis is needed for CSR (getCert) request as it persists the signed Cub ca cert in its rocksdb.
> Like I mentioned in the beginning, this is not a common case but since we already persist certs in the SCM rocksdb, a tool to restore certs should be helpful in this case.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@ozone.apache.org
For additional commands, e-mail: issues-help@ozone.apache.org