Back to bug 1895040

Who When What Removed Added
Christina Meno 2020-11-05 15:51:17 UTC Target Release 4.2 4.1
Target Milestone rc z3
Christina Meno 2020-11-05 15:59:36 UTC Status NEW ASSIGNED
Assignee nojha eshiskki
Drew Harris 2020-11-09 16:45:14 UTC Target Release 4.1 4.1z3
CC anharris
Target Milestone z3 ---
Christina Meno 2020-11-10 22:10:27 UTC Status ASSIGNED POST
Flags needinfo?(eshiskki)
Ken Dreyer (Red Hat) 2020-11-11 00:36:06 UTC Status POST MODIFIED
Fixed In Version ceph-14.2.8-113.el8cp, ceph-14.2.8-113.el7cp
errata-xmlrpc 2020-11-12 03:57:21 UTC Status MODIFIED ON_QA
errata-xmlrpc 2020-12-02 09:55:42 UTC Flags needinfo?(eshiskki)
Status ON_QA RELEASE_PENDING
errata-xmlrpc 2020-12-02 15:22:34 UTC Status RELEASE_PENDING CLOSED
Resolution --- ERRATA
Last Closed 2020-12-02 15:22:34 UTC
errata-xmlrpc 2020-12-02 15:22:53 UTC Link ID Red Hat Product Errata RHSA-2020:5325
Aron Gunn 2021-01-28 21:04:54 UTC CC agunn
Docs Contact agunn
Flags needinfo?(eshiskki)
Aron Gunn 2021-02-02 15:59:32 UTC Flags needinfo?(eshiskki)
Flags needinfo?(eshiskki)
Aron Gunn 2021-02-03 20:23:40 UTC Doc Text Cause: A number of endian bugs in procedures of encoding/decoding of Ceph messages

Consequence: 1) Deployment of a hybrid ceph cluster built of nodes over different architectures (x86_64 and s390) crashes. 2) Monitor daemon of homogeneous cluster
Doc Text built of s390x nodes crashes repeatedly under specific workload on Ceph FS.


Fix: 1) Decode entity_addrvec_t with marker 1 properly - exactly by the same way as done by entity_addr_t::decode.
2) Decode enum types on big-endian systems properly
Doc Text , using an intermediate variable of the integer type. 3) encode/decode float types on big-endian systems properly.

Result: Deployment of hybrid Ceph cluster is going as expected. Monitor daemon of a homogeneous cluster built of s390x nodes doesn't
Doc Text crash.
Doc Type If docs needed, set a value Bug Fix
Flags needinfo?(eshiskki)
Doc Text Cause: A number of endian bugs in procedures of encoding/decoding of Ceph messages

Consequence: 1) Deployment of a hybrid ceph cluster built of nodes over different architectures (x86_64 and s390) crashes. 2) Monitor daemon of homogeneous cluster built of s390x nodes crashes repeatedly under specific workload on Ceph FS.


Fix: 1) Decode entity_addrvec_t with marker 1 properly - exactly by the same way as done by entity_addr_t::decode.
2) Decode enum types on big-endian systems properly, using an intermediate variable of the integer type. 3) encode/decode float types on big-endian systems properly.

Result: Deployment of hybrid Ceph cluster is going as expected. Monitor daemon of a homogeneous cluster built of s390x nodes doesn't crash.
.Improvements to the encoding and decoding of messages on storage clusters

When deploying a {storage-product} cluster containing a heterogeneous architecture, such as x86_64 and s390, could cause system crashes. Also, under certain workloads for CephFS, Ceph Monitors on s390x nodes could crash unexpectedly. With this release, decoding `entity_addrvec_t` with a marker of 1 properly, decode `enum` types on big-endian systems properly by using an intermediate integer variable type, and fixed encoding and decoding `float` types on big-endian systems. As a result, heterogeneous storage clusters, and Ceph Monitors on s390x nodes no longer crash.
Aron Gunn 2021-02-03 20:24:27 UTC Blocks 1816167
Aron Gunn 2021-02-03 20:36:11 UTC Doc Text .Improvements to the encoding and decoding of messages on storage clusters

When deploying a {storage-product} cluster containing a heterogeneous architecture, such as x86_64 and s390, could cause system crashes. Also, under certain workloads for CephFS, Ceph Monitors on s390x nodes could crash unexpectedly. With this release, decoding `entity_addrvec_t` with a marker of 1 properly, decode `enum` types on big-endian systems properly by using an intermediate integer variable type, and fixed encoding and decoding `float` types on big-endian systems. As a result, heterogeneous storage clusters, and Ceph Monitors on s390x nodes no longer crash.
.Improvements to the encoding and decoding of messages on storage clusters

When deploying a {storage-product} cluster containing a heterogeneous architecture, such as x86_64 and s390, could cause system crashes. Also, under certain workloads for CephFS, Ceph Monitors on s390x nodes could crash unexpectedly. With this release, properly decoding `entity_addrvec_t` with a marker of `1`, properly decoding the `enum` types on big-endian systems by using an intermediate integer variable type, and fixed encoding and decoding `float` types on big-endian systems. As a result, heterogeneous storage clusters, and Ceph Monitors on s390x nodes no longer crash.

Back to bug 1895040