[
https://issues.apache.org/jira/browse/HBASE-25441?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
lujie updated HBASE-25441:
--------------------------
Description:
||API||Severity||symptom||
|clearRegionBlockCache|Severe|The API will call
LruBlockCache.evictBlocksByHfileName,
who is declared as an expensive operation(see its comments), thus non-amin may
result Dos|
|clearSlowLogsResponses|Normal|clears queue records from ringbuffer|
|updateConfiguration|Normal|non-admin user can make RS reload configutation
from disk by this API. |
|updateRegionFavoredNodesMapping|Normal|Non-admin user can change the region's
best storage location by this api|
|stopServer|low|stopServer on RS is slient, which make client think he/she
success shutdown RS.
Add preRpcCheck ont only make client receive the failed message,
but also prevent the non-admin user stop the RS,
even the hbase.coprocessor.regionserver.classes are not configured.|
was:
||API||Severity||symptom||
|clearRegionBlockCache|Severe|The API will call
LruBlockCache.evictBlocksByHfileName,
who is declared as an expensive operation(see its comments), thus non-amin may
result Dos|
|clearSlowLogsResponses|Normal|clears queue records from ringbuffer|
|updateConfiguration|Normal|non-admin user can make RS reload configutation
from disk by this API. Only admin should be allowed to reconfig a cluster(see
ZOOKEEPER-2014|
|updateRegionFavoredNodesMapping|Normal|Non-admin user can change the region's
best storage location by this api|
|stopServer|low|stopServer on RS is slient, which make client think he/she
success shutdown RS.
Add preRpcCheck ont only make client receive the failed message,
but also prevent the non-admin user stop the RS,
even the hbase.coprocessor.regionserver.classes are not configured.|
> add security check for some APIs in RSRpcServices
> -------------------------------------------------
>
> Key: HBASE-25441
> URL: https://issues.apache.org/jira/browse/HBASE-25441
> Project: HBase
> Issue Type: Bug
> Reporter: lujie
> Priority: Critical
>
>
> ||API||Severity||symptom||
> |clearRegionBlockCache|Severe|The API will call
> LruBlockCache.evictBlocksByHfileName,
> who is declared as an expensive operation(see its comments), thus non-amin
> may result Dos|
> |clearSlowLogsResponses|Normal|clears queue records from ringbuffer|
> |updateConfiguration|Normal|non-admin user can make RS reload configutation
> from disk by this API. |
> |updateRegionFavoredNodesMapping|Normal|Non-admin user can change the
> region's best storage location by this api|
> |stopServer|low|stopServer on RS is slient, which make client think he/she
> success shutdown RS.
> Add preRpcCheck ont only make client receive the failed message,
> but also prevent the non-admin user stop the RS,
> even the hbase.coprocessor.regionserver.classes are not configured.|
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
