Documentation + more work on restore -- kes12Aug02

[bacula/bacula] / bacula / kernstodo

                 Kern's ToDo List
                  11 August 2002

Irix conversion notes:
- no uuencode
- no hostname
To do:    
- Document passwords.
- Document running multiple Jobs
- Document that two Verifys at same time on same client do not work.
- Document how to recycle a tape in 7 days even if the backup takes a long time.
- Document default config file locations.
- Document better includes (does it cross file systems ?).
- Document specifically how to add new File daemon to config files.
- Document forcing a new tape to be used.
- Document "Error in message.c:500 Mail program terminated in error.

From Chuck:
--bindir is wrong and does not reflect prefix= in the *_sqlite_* scripts
  (src/cats)
--top level configure options are not passed to the depkgs, particularly
  prefix=
--Also, it might be better to split the depkgs location from the --with-sqlite
  location. 
--should be able to specify e.g. --with-sqlite=/opt/local and have it find
  lib, bin, sbin for itself
  I tried this and it didn't find sqlite.h
--sd.conf password does not match dir.conf storage password
=======

- Fix catalog filename truncation in sql_get and elsewhere. Use
  only a single filename split routine.
- Add command to reset VolFiles to a larger value (don't allow
  a smaller number or print big warning).
- Make SD disallow writing on Volume with fewer files than in
  the catalog.
- Make Restore report an error if FD or SD term codes are not OK.
- Convert all %x substitution variables, which are hard to remember
  and read to %(variable-name)s.  Idea from TMDA.
- Report volume write rate.
- Fix db_get_job_volume_names() to return array of strings.
- Report compression % and other compression statistics if turned on.
- Add JobLevel in FD status (but make sure it is defined).
- Eliminate MySQL shared libraries from smtp and daemons not
  using MySQL.
- Pass "Catalog Files = no" to storage daemon to eliminate
  network traffic.
- Make Pool resource handle Counter resources.
- Remove NextId for SQLite. Optimize.
- Fix gethostbyname() to use gethostbyname_r()
- Cleanup path/filename separation in sql_get.c and sql_create.c
- Implement ./configure --with-client-only
- Strip trailing / from Include
- Move all SQL statements into a single location.
- Cleanup db_update_media and db_update_pool
- Add UA rc and history files.
- put termcap (used by console) in ./configure and
  allow -with-termcap-dir.
- Remove JobMediaId it is not used.
- Enhance time and size scanning routines.
- Fix Autoprune for Volumes to respect need for full save.
- DateWritten may be wrong.
- Fix Win32 config file definition name on /install
- When we are at EOM, we must ask each job to write JobMedia
  record (update_volume_info).
- No READLINE_SRC if found in alternate directory.
- Add Client FS/OS id (Linux, Win95/98, ...).
- Put Windows files in Windows stream?

====== 31 May 2002 ========
Now that Bacula 1.20 is released, virtually all the 
basic features are implemented (some are still quite
primitive though). Over the next month or two, I'm 
planning to focus on the following items:

Minor details:
- Fix any bugs I find or you report.
- Finish the implementation of automatic pruning 
  (add pruning of Restore and Verify jobs). 
- Make sure pruning of Volumes won't prune the
  only backup of a FileSet

Major Project:
- Improve the Restore capabilities of Bacula
  - Restore to most recent system state (i.e.
    figure out what tapes need to be mounted and
    in what order).
  - Restore to a particular time (perhaps several
    variations -- e.g. before date, after date).
  - Interactive Restore where you get to select 
    what files are to be restored (much like the Unix
    "restore" program permits). Now that we have a
    catalog of all files saved, it would be nice to
    be able to use it. 
  - Restore options (overwrite, overwrite if older,
    overwrite if newer, never overwrite, ...)
  - Improve the standalone programs (bls and bextract)
    to have pattern matching capabilities (e.g. restore
    by FileSet, Job, JobType, JobLevel, ...). 
  - Ideally after each Job, Bacula could write out a 
    set of commands to a file that if later feed to
    bextract would restore your system to the current
    state (at least for the saved FileSet).  This would 
    provide a simple disaster recovery that could be 
    initiated from a "floppy" and one simple ASCII control 
    file. I'm not exactly sure how to do this, but it
    shouldn't be too hard and I'll 
    be trying to go in this direction.
  
Smaller Projects:
- Implement tape verification to ensure that the data
  written for a particular Job can really be read.
- Compare tape File attributes to Catalog.
  (File attributes are size, dates, MD5, but not
   data).
- Compare tape to Client files (attributes, or
  attributes and data) 

Playing around:
- With the current Bacula 1.21 (not yet in the CVS) I
  expect there is about 95% chance that running multiple
  simultaneous Jobs will actually work without stepping
  on each other. I'm planning to try this sometime soon.
===========

Projects:
- Add Base job.
- Rework Storage daemon with new rwl_lock routines.
- Implement Label templates
- Pass JCR to database routines permitting better error printing.
- Improve Restore
- Verify tape data
- Verify against Full.

Dump:
  mysqldump -f  --opt bacula >bacula


To be done:
- Probably add End of Data tape records (this would make
  the tape format incompatible with the previous version).
- I'll most likely enhance the current tape format
  in the way that I previously described, which will make
  some of the labels incompatible, but the change will
  not affect the current restore code since it does not
  look at the details of the labels.
- I may add a few more waiting conditions in the Storage
  daemon where it will current immediately aborts a
  Job if the necessary resources are not available (e.g.
  tape is being written and a read request arrives).
- Write an applet for Linux.

- Remove PoolId from Job table, it exists in Media.
- Allow commands to detach or run in background.
- Write better dump of Messages resource.
- Fix status delay on storage daemon during rewind.
- Add VerNo to each Session label record. 
- Add Job to Session records.
- Add VOLUME_CAT_INFO to the EOS tape record (as
  well as to the EOD record).
- Add SD message variables to control operator wait time
  - Maximum Operator Wait
  - Minimum Message Interval
  - Maximum Message Interval
- Add EOM handling variables
  - Write EOD records
  - Require EOD records
- Send Operator message when cannot read tape label.
- Think about how to handle I/O error on MTEOM.
- If Storage daemon aborts a job, ensure that this
  is printed in the error message.
- Verify level=Volume (scan only), level=Data (compare of data to file).
  Verify level=Catalog, level=InitCatalog
- Scan tape contents into database.
- Dump of Catalog
- Cold start full restore (restore catalog then
  user selects what to restore).  Write summary file containing only
  Job, Media, and Catalog information. Store on another machine.
- Dump/Restore database
- File system type
- Events file
- Implement first cut of Catalog Retention period (remove old
  entries from database).
- Add SessionTime/Id filters to bextract.
- Write bscan
- Ensure that Start/End File/Block are correct.
- Add keyword search to show command in Console.
- If MySQL database is not running, job terminates with
  wierd type and wierd error code.
- Write a regression script
- Report bad status from smtp or mail program.
- Fix Win2000 error with no messages during startup.
- Add estimate to Console
- Events : tape has more than xxx bytes.
- In Storage daemon, status should include job cancelled.
- Write general list maintenance subroutines.
- Implement immortal format with EDOs. 
- Restrict characters permitted in a Resource name.
- Restore file xx or files xx, yy to their most recent values.
- Provide definitive identification of  type in backup.
- Complete  code in Bacula Resources -- this will permit
  reading a new config file at any time.
- Document new Console
- Handle ctl-c in Console
- Test restore of Windows backup
- Implement LabelTemplate (at least first cut).
- Implement script driven addition of File daemon to
  config files.

- Bug: anonymous Volumes requires mount in some cases.
- see setgroup and user for Bacula p4-5 of stunnel.c
- Implement new serialize subroutines
   send(socket, "string", &Vol, "uint32", &i, NULL)
- Add save type to Session label.
- Correct date on Session label.
- On I/O error, write EOF, then try to write again.
- Audit all UA commands to ensure that we always prompt where
  possible.
- If ./btape is called without /dev, assume argument is
  a Storage resource name.
- Put memory utilization in Status output of each daemon
  if full status requested or if some level of debug on.
- Make database type selectable by .conf files i.e. at runtime
- gethostbyname failure in bnet_connect() continues
  generating errors -- should stop.
- Don't create a volume that is already written. I.e. create only once.
- If error at end of tape, implement some way to kill waiting processes.
- Get correct block/file information in Catalog, pay attention
  to change of media.
- Add HOST to Volume label.
- Set flag for uname -a.  Add to Volume label.
- Implement throttled work queue.
- Write bscan program that will syncronize the DB Media record with
  the contents of the Volume -- for use after a crash.
- Check for EOT at ENOSPC or EIO or ENXIO (unix Pc)
- Allow multiple Storage specifications (or multiple names on
  a single Storage specification) in the Job record. Thus a job 
  can be backed up to a number of storage devices.
- Implement full MediaLabel code.
- Implement dump label to UA
- Copy volume using single drive.
- Copy volume with multiple driven (same or different block size).     
- Add block size (min, max) to Vol label.
- Concept of VolumeSet during restore which is a list
  of Volume names needed.
- Restore files modified after date
- Restore file modified before date
- Emergency restore info:
  - Backup Bacula
  - Backup working directory
  - Backup Catalog
- Restore options (do not overwrite)
- Restore -- do nothing but show what would happend
- Authentication between SD and FD
- SET LD_RUN_PATH=$HOME/mysql/lib/mysql
- Send Volumes needed during restore to Console
- Put Job statistics in End Session Label (files saved,
  total bytes, start time, ...).     
- Put FileSet name in the SOS label.
- Implement Restore FileSet=
- Write a scanner for the UA (keyword, scan-routine, result, prompt).
- Create a protocol.h and protocol.c where all protocol messages
  are concentrated.
- If SD cannot open a drive, make it periodically retry.
- Put Bacula version somewhere in Job stream, probably Start Session
  Labels.
- Remove duplicate fields from jcr (e.g. jcr.level and
  jcr.jr.Level, ...).
- Timout a job or terminate if link goes down, or reopen link and query.
- Define how we handle times to avoid problem with Unix dates (2049 ?).
- The daemons should know when one is already
  running and refuse to run a second copy.
- Fill all fields in Vol/Job Header -- ensure that everything
  needed is written to tape. Think about restore to Catalog
  from tape.  Client record needs improving.
- Find general solution for sscanf size problems (as well
  as sprintf. Do at run time?

- Concept of precious tapes (cannot be reused).
- Allow FD to run from inetd ???
- Preprocessing command per file.
- Postprocessing command per file (when restoring).

- Restore should get Device and Pool information from
  job record rather than from config.
- Make SD send attribute stream to DR but first
  buffering to file, then sending only when the
  files are written to tape.
- Autolabel should be specified by DR instead of SD.
- Ability to recreate the catalog from a tape.
- Find out how to get the system tape block limits, e.g.:
  Apr 22 21:22:10 polymatou kernel: st1: Block limits 1 - 245760 bytes.  
  Apr 22 21:22:10 polymatou kernel: st0: Block limits 2 - 16777214 bytes.
- Storage daemon    
  - Add media capacity
  - AutoScan (check checksum of tape)
  - Format command = "format /dev/nst0"
  - MaxRewindTime
  - MinRewindTime
  - MaxBufferSize
  - Seek resolution (usually corresponds to buffer size)
  - EODErrorCode=ENOSPC or code
  - Partial Read error code
  - Partial write error code
  - Nonformatted read error
  - Nonformatted write error
  - WriteProtected error
  - IOTimeout
  - OpenRetries
  - OpenTimeout
  - IgnoreCloseErrors=yes
  - Tape=yes
  - NoRewind=yes
- Pool
  - Maxwrites
  - Recycle period
- Job
  - MaxWarnings
  - MaxErrors (job?)
=====
- Eliminate duplicate File records to shrink database.
- FD sends unsaved file list to Director at end of job. 
- Implement InsertUniqueDB. 
- Write a Storage daemon that uses pipes and
  standard Unix programs to write to the tape.
  See afbackup.
- Need something that monitors the JCR queue and
  times out jobs by asking the deamons where they are.
- Add daemon JCR JobId=0 to have a daemon context
- Pool resource
  - Auto label
  - Auto media verify
  - Client (list of clients to force client)
  - Devices (list of devices to force device)
  - enable/disable
  - Groups
  - Levels
  - Type: Backup, ...
  - Recycle from other pools: Yes, No
  - Recycle to other pools: Yes, no
  - FileSets
  - MaxBytes?
  - Optional MediaType to force media?
  - Maintain Catalog
  - Label Template
  - Retention Period
  ============
  - Name
  - NumVols
  - NaxVols
  - CurrentVol

=====
  if(connect(sockfd, (struct sockaddr * ) (& addr), sizeof(addr)) .lt. 0){
    close(sockfd);
    return(-6);
  }

  linger.l_onoff = 1;
  linger.l_linger = 60;
  i = setsockopt(sockfd, SOL_SOCKET, SO_LINGER, (char *) &linger,
                                                sizeof (linger));

  fl = fcntl(sockfd, F_GETFL);
  fcntl(sockfd, F_SETFL, fl & (~ O_NONBLOCK) & (~ O_NDELAY));
====
- Add "0nnn" in front of all sscanf %s fields
  to prevent field overflow.
- Restore:
  What: jobid or file list
  From: tape, file, ...
  Where: original location, another path
  How: Always replace, Replace if newer, Never replace
  Report: files restored; files not restored; errors; warnings
          summary.
- Enhance Jmsg code to permit buffering and saving to disk.
- Probably create a jcr with JobId=0 as a master
  catchall if jcr not found or if operation involves
  global operation.
- device driver = "xxxx" for drives.
- restart: paranoid: read label fsf to
  eom read append block, and go
  super-paranoid: read label, read all files
  in between, read append block, and go
  verify: backspace, read append block, and go
  permissive: same as above but frees drive
  if tape is not valid.
- Verify from Volume
- Ensure that /dev/null works
- File daemon should build list of files skipped, and then
  at end of save retry and report any errors.
- Need report class for messages. Perhaps
  report resource where report=group of messages
- Extract what=(session_id|file_list); where
- Verify from Tape
- enhance scan_attrib and rename scan_jobtype, and
  fill in code for "since" option 
- dir_config: get rid of all printfs
- To buffer messages, we need associated jobid and Director name.
- Need to save contents of FileSet to tape?
- Director needs a time after which the report status is sent
  anyway -- or better yet, a retry time for the job.
  Don't reschedule a job if previous incarnation is still running.
- Figure out how to do a "full" restore from catalog
- Figure out how to save the catalog (possibly a special FileSet).
- Figure out how to restore the catalog.
- Figure out how to put a Volume into the catalog (from the tape)
- Figure out how to do a restore from a Volume
- Some way to automatically backup everything is needed????
- Need a structure for pending actions:
  - buffered messages
  - termination status (part of buffered msgs?)
- Concept of grouping Storage devices and job can use
  any of a number of devices
- Drive management
  Read, Write, Clean, Delete
- Login to Bacula; Bacula users with different permissions:
   owner, group, user
- Tape recycle destination
- Job Schedule Status
  - Automatic
  - Manual
  - Running
- File daemon should pass Director the operating system info
  to be stored in the Client Record (or verified that it has
  not changed).
- Store info on each file system type (probably in the job header on tape.
  This could be the output of df; or perhaps some sort of /etc/mtab record.

Longer term to do:
- Use media 1 time (so that we can do 6 days of incremental
  backups before switching to another tape) (already)
  specify # times (jobs)
  specify bytes (already)
  specify time (seconds, hours, days)
- Implement FSM (File System Modules).
- Identify unchanged or "system" files and save them to a
  special tape thus removing them from the standard 
  backup FileSet -- BASE backup.
- Turn virutally all sprintfs into snprintfs.
- Heartbeat between daemons.
- Audit M_ error codes to ensure they are correct and
  consistent.
- Add variable break characters to lex analyzer.
  Either a bit mask or a string of chars so that
  the caller can change the break characters.
- Make a single T_BREAK to replace T_COMMA, etc.
- Ensure that File daemon and Storage daemon can
  continue a save if the Director goes down (this
  is NOT currently the case). Must detect socket error,
  buffer messages for later. 


Done: (see kernsdone for more)
--the console script is broken as installed and has to be hand-massaged with
  paths, config files etc.
- Termination status in FD for Verify = C -- incorrect.
- Implement alter_sqlite_tables
- Fix scheduler -- see "Hourly cycle". It doesn't do both each
  hour, rather it alternates between 0:05 and 0:35.
- Create Counter DB records.