Chapter 8. Troubleshooting objects
As a storage administrator, you can use the ceph-objectstore-tool
utility to perform high-level or low-level object operations. The ceph-objectstore-tool
utility can help you troubleshoot problems related to objects within a particular OSD or placement group.
Manipulating objects can cause unrecoverable data loss. Contact Red Hat support before using the ceph-objectstore-tool
utility.
8.1. Prerequisites
- Verify there are no network related issues.
8.2. Troubleshooting high-level object operations
As a storage administrator, you can use the ceph-objectstore-tool
utility to perform high-level object operations. The ceph-objectstore-tool
utility supports the following high-level object operations:
- List objects
- List lost objects
- Fix lost objects
Manipulating objects can cause unrecoverable data loss. Contact Red Hat support before using the ceph-objectstore-tool
utility.
8.2.1. Prerequisites
-
Having
root
access to the Ceph OSD nodes.
8.2.2. Listing objects
The OSD can contain zero to many placement groups, and zero to many objects within a placement group (PG). The ceph-objectstore-tool
utility allows you to list objects stored within an OSD.
Prerequisites
-
Having
root
access to the Ceph OSD node. -
Stopping the
ceph-osd
daemon.
Procedure
Verify the appropriate OSD is down:
Syntax
systemctl status ceph-osd@$OSD_NUMBER
Example
[root@osd ~]# systemctl status ceph-osd@1
Identify all the objects within an OSD, regardless of their placement group:
Syntax
ceph-objectstore-tool --data-path $PATH_TO_OSD --op list
Example
[root@osd ~]# ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-0 --op list
Identify all the objects within a placement group:
Syntax
ceph-objectstore-tool --data-path $PATH_TO_OSD --pgid $PG_ID --op list
Example
[root@osd ~]# ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-0 --pgid 0.1c --op list
Identify the PG an object belongs to:
Syntax
ceph-objectstore-tool --data-path $PATH_TO_OSD --op list $OBJECT_ID
Example
[root@osd ~]# ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-0 --op list default.region
Additional Resources
- For more information on stopping an OSD, see the Starting, Stopping, and Restarting a Ceph Daemons by Instance section in the Red Hat Ceph Storage Administration Guide.
8.2.3. Fixing lost objects
You can use the ceph-objectstore-tool
utility to list and fix lost and unfound objects stored within a Ceph OSD. This procedure applies only to legacy objects.
Prerequisites
-
Having
root
access to the Ceph OSD node. -
Stopping the
ceph-osd
daemon.
Procedure
Verify the appropriate OSD is down:
Syntax
systemctl status ceph-osd@$OSD_NUMBER
Example
[root@osd ~]# systemctl status ceph-osd@1
To list all the lost legacy objects:
Syntax
ceph-objectstore-tool --data-path $PATH_TO_OSD --op fix-lost --dry-run
Example
[root@osd ~]# ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-0 --op fix-lost --dry-run
Use the
ceph-objectstore-tool
utility to fix lost and unfound objects. Select the appropriate circumstance:To fix all lost objects:
Syntax
ceph-objectstore-tool --data-path $PATH_TO_OSD --op fix-lost
Example
[root@osd ~]# ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-0 --op fix-lost
To fix all the lost objects within a placement group:
Syntax
ceph-objectstore-tool --data-path $PATH_TO_OSD --pgid $PG_ID --op fix-lost
Example
[root@osd ~]# ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-0 --pgid 0.1c --op fix-lost
To fix a lost object by its identifier:
Syntax
ceph-objectstore-tool --data-path $PATH_TO_OSD --op fix-lost $OBJECT_ID
Example
[root@osd ~]# ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-0 --op fix-lost default.region
Additional Resources
- For more information on stopping an OSD, see the Starting, Stopping, and Restarting a Ceph Daemons by Instance section in the Red Hat Ceph Storage Administration Guide.
8.3. Troubleshooting low-level object operations
As a storage administrator, you can use the ceph-objectstore-tool
utility to perform low-level object operations. The ceph-objectstore-tool
utility supports the following low-level object operations:
- Manipulate the object’s content
- Remove an object
- List the object map (OMAP)
- Manipulate the OMAP header
- Manipulate the OMAP key
- List the object’s attributes
- Manipulate the object’s attribute key
Manipulating objects can cause unrecoverable data loss. Contact Red Hat support before using the ceph-objectstore-tool
utility.
8.3.1. Prerequisites
-
Having
root
access to the Ceph OSD nodes.
8.3.2. Manipulating the object’s content
With the ceph-objectstore-tool
utility, you can get or set bytes on an object.
Setting the bytes on an object can cause unrecoverable data loss. To prevent data loss, make a backup copy of the object.
Prerequisites
-
Having
root
access to the Ceph OSD node. -
Stopping the
ceph-osd
daemon.
Procedure
Verify the appropriate OSD is down:
Syntax
systemctl status ceph-osd@$OSD_NUMBER
Example
[root@osd ~]# systemctl status ceph-osd@1
- Find the object by listing the objects of the OSD or placement group (PG).
Before setting the bytes on an object, make a backup and a working copy of the object:
Syntax
ceph-objectstore-tool --data-path $PATH_TO_OSD --pgid $PG_ID \ $OBJECT \ get-bytes > $OBJECT_FILE_NAME ceph-objectstore-tool --data-path $PATH_TO_OSD --pgid $PG_ID \ $OBJECT \ get-bytes > $OBJECT_FILE_NAME
Example
[root@osd ~]# ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-0 --pgid 0.1c \ '{"oid":"zone_info.default","key":"","snapid":-2,"hash":235010478,"max":0,"pool":11,"namespace":""}' \ get-bytes > zone_info.default.backup [root@osd ~]# ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-0 --pgid 0.1c \ '{"oid":"zone_info.default","key":"","snapid":-2,"hash":235010478,"max":0,"pool":11,"namespace":""}' \ get-bytes > zone_info.default.working-copy
- Edit the working copy object file and modify the object contents accordingly.
Set the bytes of the object:
Syntax
ceph-objectstore-tool --data-path $PATH_TO_OSD --pgid $PG_ID \ $OBJECT \ set-bytes < $OBJECT_FILE_NAME
Example
[root@osd ~]# ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-0 --pgid 0.1c \ '{"oid":"zone_info.default","key":"","snapid":-2,"hash":235010478,"max":0,"pool":11,"namespace":""}' \ set-bytes < zone_info.default.working-copy
Additional Resources
- For more information on stopping an OSD, see the Starting, Stopping, and Restarting a Ceph Daemons by Instance section in the Red Hat Ceph Storage Administration Guide.
8.3.3. Removing an object
Use the ceph-objectstore-tool
utility to remove an object. By removing an object, its contents and references are removed from the placement group (PG).
You cannot recreate an object once it is removed.
Prerequisites
-
Having
root
access to the Ceph OSD node. -
Stopping the
ceph-osd
daemon.
Procedure
Remove an object:
Syntax
ceph-objectstore-tool --data-path $PATH_TO_OSD --pgid $PG_ID \ $OBJECT \ remove
Example
[root@osd ~]# ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-0 --pgid 0.1c \ '{"oid":"zone_info.default","key":"","snapid":-2,"hash":235010478,"max":0,"pool":11,"namespace":""}' \ remove
Additional Resources
- For more information on stopping an OSD, see the Starting, Stopping, and Restarting a Ceph Daemons by Instance section in the Red Hat Ceph Storage Administration Guide.
8.3.4. Listing the object map
Use the ceph-objectstore-tool
utility to list the contents of the object map (OMAP). The output provides you a list of keys.
Prerequisites
-
Having
root
access to the Ceph OSD node. -
Stopping the
ceph-osd
daemon.
Procedure
Verify the appropriate OSD is down:
Syntax
systemctl status ceph-osd@$OSD_NUMBER
Example
[root@osd ~]# systemctl status ceph-osd@1
List the object map:
Syntax
ceph-objectstore-tool --data-path $PATH_TO_OSD --pgid $PG_ID \ $OBJECT \ list-omap
Example
[root@osd ~]# ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-0 --pgid 0.1c \ '{"oid":"zone_info.default","key":"","snapid":-2,"hash":235010478,"max":0,"pool":11,"namespace":""}' \ list-omap
Additional Resources
- For more information on stopping an OSD, see the Starting, Stopping, and Restarting a Ceph Daemons by Instance section in the Red Hat Ceph Storage Administration Guide.
8.3.5. Manipulating the object map header
The ceph-objectstore-tool
utility will output the object map (OMAP) header with the values associated with the object’s keys.
If using FileStore as the OSD backend object store, then add the --journal-path $PATH_TO_JOURNAL
argument when getting or setting the object map header. Where the $PATH_TO_JOURNAL
variable is the absolute path to the OSD journal, for example /var/lib/ceph/osd/ceph-0/journal
.
Prerequisites
-
Having
root
access to the Ceph OSD node. -
Stopping the
ceph-osd
daemon.
Procedure
Verify the appropriate OSD is down:
Syntax
systemctl status ceph-osd@$OSD_NUMBER
Example
[root@osd ~]# systemctl status ceph-osd@1
Get the object map header:
Syntax
ceph-objectstore-tool --data-path $PATH_TO_OSD \ --pgid $PG_ID $OBJECT \ get-omaphdr > $OBJECT_MAP_FILE_NAME
Example
[root@osd ~]# ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-0 \ --pgid 0.1c '{"oid":"zone_info.default","key":"","snapid":-2,"hash":235010478,"max":0,"pool":11,"namespace":""}' \ get-omaphdr > zone_info.default.omaphdr.txt
Set the object map header:
Syntax
ceph-objectstore-tool --data-path $PATH_TO_OSD \ --pgid $PG_ID $OBJECT \ get-omaphdr < $OBJECT_MAP_FILE_NAME
Example
[root@osd ~]# ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-0 \ --pgid 0.1c '{"oid":"zone_info.default","key":"","snapid":-2,"hash":235010478,"max":0,"pool":11,"namespace":""}' \ set-omaphdr < zone_info.default.omaphdr.txt
Additional Resources
- For more information on stopping an OSD, see the Starting, Stopping, and Restarting a Ceph Daemons by Instance section in the Red Hat Ceph Storage Administration Guide.
8.3.6. Manipulating the object map key
Use the ceph-objectstore-tool
utility to change the object map (OMAP) key. You need to provide the data path, the placement group identifier (PG ID), the object, and the key in the OMAP.
If using FileStore as the OSD backend object store, then add the --journal-path $PATH_TO_JOURNAL
argument when getting, setting or removing the object map key. Where the $PATH_TO_JOURNAL
variable is the absolute path to the OSD journal, for example /var/lib/ceph/osd/ceph-0/journal
.
Prerequisites
-
Having
root
access to the Ceph OSD node. -
Stopping the
ceph-osd
daemon.
Procedure
Get the object map key:
Syntax
ceph-objectstore-tool --data-path $PATH_TO_OSD \ --pgid $PG_ID $OBJECT \ get-omap $KEY > $OBJECT_MAP_FILE_NAME
Example
[root@osd ~]# ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-0 \ --pgid 0.1c '{"oid":"zone_info.default","key":"","snapid":-2,"hash":235010478,"max":0,"pool":11,"namespace":""}' \ get-omap "" > zone_info.default.omap.txt
Set the object map key:
Syntax
ceph-objectstore-tool --data-path $PATH_TO_OSD \ --pgid $PG_ID $OBJECT \ set-omap $KEY < $OBJECT_MAP_FILE_NAME
Example
[root@osd ~]# ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-0 \ --pgid 0.1c '{"oid":"zone_info.default","key":"","snapid":-2,"hash":235010478,"max":0,"pool":11,"namespace":""}' \ set-omap "" < zone_info.default.omap.txt
Remove the object map key:
Syntax
ceph-objectstore-tool --data-path $PATH_TO_OSD \ --pgid $PG_ID $OBJECT \ rm-omap $KEY
Example
[root@osd ~]# ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-0 \ --pgid 0.1c '{"oid":"zone_info.default","key":"","snapid":-2,"hash":235010478,"max":0,"pool":11,"namespace":""}' \ rm-omap ""
Additional Resources
- For more information on stopping an OSD, see the Starting, Stopping, and Restarting a Ceph Daemons by Instance section in the Red Hat Ceph Storage Administration Guide.
8.3.7. Listing the object’s attributes
Use the ceph-objectstore-tool
utility to list an object’s attributes. The output provides you with the object’s keys and values.
If using FileStore as the OSD backend object store, then add the --journal-path $PATH_TO_JOURNAL
argument when listing an object’s attributes. Where the $PATH_TO_JOURNAL
variable is the absolute path to the OSD journal, for example /var/lib/ceph/osd/ceph-0/journal
.
Prerequisites
-
Having
root
access to the Ceph OSD node. -
Stopping the
ceph-osd
daemon.
Procedure
Verify the appropriate OSD is down:
Syntax
systemctl status ceph-osd@$OSD_NUMBER
Example
[root@osd ~]# systemctl status ceph-osd@1
List the object’s attributes:
Syntax
ceph-objectstore-tool --data-path $PATH_TO_OSD \ --pgid $PG_ID $OBJECT \ list-attrs
Example
[root@osd ~]# ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-0 \ --pgid 0.1c '{"oid":"zone_info.default","key":"","snapid":-2,"hash":235010478,"max":0,"pool":11,"namespace":""}' \ list-attrs
Additional Resources
- For more information on stopping an OSD, see the Starting, Stopping, and Restarting a Ceph Daemons by Instance section in the Red Hat Ceph Storage Administration Guide.
8.3.8. Manipulating the object attribute key
Use the ceph-objectstore-tool
utility to change an object’s attributes. To manipulate the object’s attributes you need the data and journal paths, the placement group identifier (PG ID), the object, and the key in the object’s attribute.
If using FileStore as the OSD backend object store, then add the --journal-path $PATH_TO_JOURNAL
argument when getting, setting or removing the object’s attributes. Where the $PATH_TO_JOURNAL
variable is the absolute path to the OSD journal, for example /var/lib/ceph/osd/ceph-0/journal
.
Prerequisites
-
Having
root
access to the Ceph OSD node. -
Stopping the
ceph-osd
daemon.
Procedure
Verify the appropriate OSD is down:
Syntax
systemctl status ceph-osd@$OSD_NUMBER
Example
[root@osd ~]# systemctl status ceph-osd@1
Get the object’s attributes:
Syntax
ceph-objectstore-tool --data-path $PATH_TO_OSD \ --pgid $PG_ID $OBJECT \ get-attrs $KEY > $OBJECT_ATTRS_FILE_NAME
Example
[root@osd ~]# ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-0 \ --pgid 0.1c '{"oid":"zone_info.default","key":"","snapid":-2,"hash":235010478,"max":0,"pool":11,"namespace":""}' \ get-attrs "oid" > zone_info.default.attr.txt
Set an object’s attributes:
Syntax
ceph-objectstore-tool --data-path $PATH_TO_OSD \ --pgid $PG_ID $OBJECT \ set-attrs $KEY < $OBJECT_ATTRS_FILE_NAME
Example
[root@osd ~]# ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-0 \ --pgid 0.1c '{"oid":"zone_info.default","key":"","snapid":-2,"hash":235010478,"max":0,"pool":11,"namespace":""}' \ set-attrs "oid" < zone_info.default.attr.txt
Remove an object’s attributes:
Syntax
ceph-objectstore-tool --data-path $PATH_TO_OSD \ --pgid $PG_ID $OBJECT \ rm-attrs $KEY
Example
[root@osd ~]# ceph-objectstore-tool --data-path /var/lib/ceph/osd/ceph-0 \ --pgid 0.1c '{"oid":"zone_info.default","key":"","snapid":-2,"hash":235010478,"max":0,"pool":11,"namespace":""}' \ rm-attrs "oid"
Additional Resources
- For more information on stopping an OSD, see the Starting, Stopping, and Restarting a Ceph Daemons by Instance section in the Red Hat Ceph Storage Administration Guide.
8.4. Additional Resources
- For Red Hat Ceph Storage support, see the Red Hat Customer Portal.