BACKUP_KILL(8)

πŸ’… BACKUP_KILL β€” the one where you say "okay bestie, we're ending this" ✨

So this is backup_kill, part of section 8 of the AFS command reference (that's the sysadmin-stuff section, iykyk), and it does exactly one thing: it ends a Backup System operation. Pending? Running? Doesn't matter, we're wrapping it up. Grab a snack, this one's got lore. πŸ“


✨ NAME ✨

backup_kill β€” Terminates a pending or running operation


πŸ¦‹ SYNOPSIS πŸ¦‹

backup kill -id <job ID or dump set name> [-help]

backup k k -i <job ID or dump set name> [-h]

πŸ’– DESCRIPTION πŸ’–

The backup kill command dequeues a Backup System operation that's pending, or terminates an operation that's running, in the current interactive session. Tiny little detail, bestie: it's available only in interactive mode. Also β€” if whoever launched the backup interactive command included the -localauth flag, or the -cell argument, or both, then those settings apply to this command too. Like they came with the group chat. πŸ’…

To terminate a dump operation, specify either the dump name (volume_set_name.dump_level_name) or its job ID number, which shows up in the output of the backup jobs command. To terminate any other type of operation, give the job ID number. That's it, that's the ticket. πŸŽ€

The effect of terminating an operation depends on the type and current state of the operation:

  • If an operation is still pending, the Tape Coordinator removes it from the queue with no other lasting effects. Clean. ✨
  • If the Tape Coordinator is unable to process the termination signal before an operation completes, it simply confirms the operation's completion. Then the operator has to take the action necessary to undo the effects of the incorrect operation. Not the vibe we wanted, but ok.
  • If a tape labeling operation is running, the effect depends on when the Tape Coordinator receives the termination signal. The labeling operation is atomic, so it either completes or does not begin at all. Use the backup readlabel command to determine if the labeling operation completed, and reissue the backup labeltape command to overwrite the incorrect label if necessary.
  • If a tape scanning operation is running, it terminates with no other effects unless the -dbadd flag was included on the backup command. In that case, the Backup System possibly has already written new Backup Database records to represent dumps on the scanned tape. If you're planning to restart the scanning operation, first locate and remove the records created during the terminated operation: a repeated backup scantape operation exits automatically when it finds that a record that it needs to create already exists. So, like, be aware. 😭
  • If a dump operation is running, all of the volumes written to the tape or backup data file before the termination signal is received are complete and usable. If the operation is restarted, the Backup System performs all the dumps again from scratch, and assigns a new dump ID number. If writing the new dumps to the same tape or file, the operator must relabel it first if the interrupted dump is not expired. If writing the new dump to a different tape or file, the operator can remove the dump record associated with the interrupted dump to free up space in the database. Iconic amounts of bookkeeping, ngl.
  • If a restore operation is running, completely restored volumes are online and usable. However β€” and this is the part where we all gasp β€” it is unlikely that many volumes are completely restored, given that complete restoration usually requires data from multiple tapes. If the termination signal comes before the Backup System has accessed all of the necessary tapes, each volume is only partially written and is never brought online. It's best to restart the restore operation from scratch to avoid possible inconsistencies. See also the "CAUTIONS" section, bestie, I'm begging you.

🚨 CAUTIONS 🚨

It is best not to issue the backup kill command against restore operations. Say it with me. πŸ™ˆ

If the termination signal interrupts a restore operation as the Backup System is overwriting an existing volume, it is possible to lose the volume entirely β€” that is, to lose both the contents of the volume as it was before the restore and any data that was restored before the termination signal arrived. The data being restored still exists on the tape, but some data can be lost permanently.

πŸ’€ fr fr, permanent data loss. This is the "do not pass go" of backup commands. Lowkey the main character of this whole page is the caution, ngl.

πŸŽ€ OPTIONS πŸŽ€

  • -id <job ID or dump set name> β€” Identifies the backup operation to terminate. Provide one of two types of values:
    • The operation's job ID number, as displayed in the output of the backup jobs command.
    • For a dump operation, either the job ID number or a dump name of the form volume_set_name.dump_level_name, where volume_set_name is the name of the volume set being dumped and dump_level_name is the last element in the dump level pathname at which the volume set is being dumped. The dump name appears in the output of the backup jobs command along with the job ID number.
  • -help β€” Prints the online help for this command. All other valid options are ignored. (Yes, it just ignores everything else. It's giving "I'm on my own path." πŸ’…)

🌈 EXAMPLES 🌈

The following command terminates the operation with job ID 5:

backup> kill 5

The following command terminates the dump operation called "user.sunday1":

backup> kill user.sunday1

πŸ‘‘ PRIVILEGE REQUIRED πŸ‘‘

The issuer must have the privilege required to initiate the operation being cancelled. But here's the cute part: because this command can be issued only within the interactive session during which the operation was initiated, the required privilege is essentially guaranteed. Slay, no privilege escalations today. 🫢


🌸 SEE ALSO 🌸

backup(8), backup_interactive(8), backup_jobs(8)


IBM Corporation 2000. <http://www.ibm.com/> All Rights Reserved.

This documentation is covered by the IBM Public License Version 1.0. It was converted from HTML to POD by software written by Chas Williams and Russ Allbery, based on work by Alf Wachsmann and Elizabeth Cassell.


That's the whole thing, bestie β€” go forth and terminate responsibly. This page came to you from OpenAFS, dated 2026-07-27, and it's filed under BACKUP_KILL(8). Mwah. πŸ’–βœ¨

πŸ“¦ where this page comes from