PowerFlex 3.X 在 NDU 期间执行mgStorageRegion_CopyFromBuffer功能时 SDS 崩溃

 - 从 3.0 升级到 3.6

 - SDS 退出即时维护模式 （IMM）  

- 启用了动态校验和。  

SDS 崩溃并出现以下死机： 

2022/12/20 10:22:18.543129 Panic in file /data/build/workspace/ScaleIO-Common-Job/src/tgt/storage/mg_impl/mg_storage_region.c, line 4188, function mgStorageRegion_CopyFromBuffer, PID 13108.Panic Expression !(bufferSizeInBytes != ((sizeInLbs) * (512))) PANIC_ID_tgt_1517847817759.
/opt/emc/scaleio/sds/bin/sds-3.6.400.107(mosDbg_PanicPrepare+0x13a) [0x93ab8a]
/opt/emc/scaleio/sds/bin/sds-3.6.400.107(mgStorageRegion_CopyFromBuffer+0x1a7) [0x810477]
/opt/emc/scaleio/sds/bin/sds-3.6.400.107(mgStorageRegion_CopyFromCachedBuffer+0x33) [0x810873]
/opt/emc/scaleio/sds/bin/sds-3.6.400.107(storageRegion_CopyFromCachedBuffer+0xde) [0x4d3f3e]
/opt/emc/scaleio/sds/bin/sds-3.6.400.107(mgStorageRegion_ReadFromSyncBuf+0x45) [0x4d4a85]
/opt/emc/scaleio/sds/bin/sds-3.6.400.107(raidComb_ReadFromRemoteUntilEntireRegionIsAcquired+0x451) [0x74e021]
/opt/emc/scaleio/sds/bin/sds-3.6.400.107(raidComb_ReadVolumeFromRemote+0x3ae) [0x74e8ee]
/opt/emc/scaleio/sds/bin/sds-3.6.400.107(raidReverseRebuild_LoopAndCopy+0x140f) [0x75518f]
/opt/emc/scaleio/sds/bin/sds-3.6.400.107(raidReverseRebuild_Start+0xb56) [0x7583c6]
/opt/emc/scaleio/sds/bin/sds-3.6.400.107(raidSyncPool_StartJob+0x375) [0x5e3145]

在存储池上启用了动态校验和功能：

Storage Pool durpsdsc7p8pool1 (Id: 469b0f6800000007) has 7 volumes and 236.9 TB (242550 GB) free net capacity. 2.5 PB (2524 TB) volume allocation limit.
...
Inflight Checksum: Enabled
...

影响

SDS 崩溃，无法退出 IMM。如果在升级过程中，升级将无法继续。

在 SDS 有助于以下目标的存储池上禁用动态校验和：

1） 查询所有 SDS 以获取 SDS 名称/id：

[root@nestedsvm2 ~]# scli --query_all_sds
Query-all-SDS returned 3 SDS nodes.

Protection Domain 8eeacbf900000000 Name: pd1
SDS ID: ab471ceb00000002 Name: svm103 State: Connected, Joined IP: 15.15.15.103 Port: 7072 Version: 3.6.500
SDS ID: ab471cea00000001 Name: svm102 State: Connected, Joined IP: 15.15.15.102 Port: 7072 Version: 3.6.500
SDS ID: ab471ce800000000 Name: svm101 State: Connected, Joined IP: 15.15.15.101 Port: 7072 Version: 3.6.500

2） 按名称/ID 查询 SDS，并查找“Storage Pool：":

 提醒：它向其提供设备的存储池;在这种情况下，只有一个，即存储池名称“SP1”。

[root@nestedsvm2 ~]# scli --query_sds --sds_id ab471ceb00000002 | grep -i 'storage pool:'
         1: Storage Pool: sp1  inflight requests factor: 115, inflight bandwidth factor 115
                Storage Pool: sp1, Capacity: 198 GB, State: Normal
                Storage Pool: sp1, Capacity: 98 GB, State: Normal
                Storage Pool: sp1, Capacity: 98 GB, State: Normal

3） 查询存储池并查找“Inflight Checksum：":

[root@nestedsvm2 ~]# scli --query_storage_pool --storage_pool_name sp1 --protection_domain_name pd1 | grep -i 'Inflight checksum'
        Inflight Checksum: Enabled

3.a） 要禁用动态校验和，可以使用 scli 或在演示服务器中完成。要使用 scli 禁用它，请运行以下命令：

[root@nestedsvm2 ~]# scli --set_checksum_mode --protection_domain_name pd1 --storage_pool_name sp1 --disable_inflight_checksum
Checksum mode modified successfully

3.b） 在演示服务器中，导航到“存储池”， > 选择“存储池 > 修改 > 常规 > ”，取消选中“启用动态校验和” > 应用”。  

 受影响的版本

PowerFlex 3.5.x

 PowerFlex 3.6.x