09/21/87 volume_backup Known errors in the current release of volume_backup. # Associated TR's Description 94 phx20536 Problem Description: The dumper bit map segment is not copied to disk after each dump cycle. The only time the dumper bit map segment is copied to the disk label is when the PV is taken off line. A problem arises when a system experiences a crash where emergency shutdown procedures cannot be performed. The ESD is responsible for taking each PV off line. Because the packs are not updated, the dumper bit map may not represent an accurate picture of what has been dumped from the pack. The result being that after a crash, the PV can only be restored from the last time that it was shutdown. Proposed Solution: To resolve this problem, the volume dumper process should be enhanced to copy the memory image of each PV's bit map to the disk label after the PV is dumped. To implement this change a call will be added to the ring-0 code of the dumper - hc_dmpr_primitives to call a new entry in the fsout_vol ring-0 subroutine. The fsout_vol subroutine is the entry called at PV shutdown time copy the dbm_seg to the disk label. A new entry point will be added to the fsout_vol subroutine that will allow the dbm_seg to be copied to disk without freeing the memory image of the dbm_seg. 93 phx20716 retrieve_from_volume [verified] Said it couldn't recover branch but went ahead and did it anyway. . As time and resources permit this problem will be evaluated and better defined at a future date. 92 phx20715 retrieve_from_volume [verified] Makes multiple attempts to retrieve the same segment from the same tape. . As time and resources permit this problem will be evaluated and better defined at a future date. 91 phx20691 volume_retrieve [investigating] The volume_retrieve command does not properly retrieve segments if missing parent directories need to be retrieved. . As time and resources permit this problem will be evaluated and better defined at a future date. 90 phx20000 retrievals [investigating] Retrieve dies if certain i/o switches are not closed. . As time and resources permit this problem will be evaluated and better defined at a future date. 89 phx17264 Volume Reloader [verified] Unable to use the "-input_volume_desc" control argument for ring 1 volume reloads. . As time and resources permit this problem will be evaluated and better defined at a future date. 88 phx20537 Volume Backup [investigating] Volume retriever takes vtoc error under certain circumstances.. As time and resources permit this problem will be evaluated and better defined at a future date. 87 phx20513 Volume Backup [investigating] The RPV pvid field in the volog headers are not being set. As time and resources permit this problem will be evaluated and better defined at a future date. 86 phx20497 Volume Retriever [investigating] If a user requests the retrieval of a file >A>B>C>x via the err command such that x doesn't currently exist online and C is a segment, the Volume Retriever will go to level 2 on the bad_dir_ condition.. As time and resources permit this problem will be evaluated and better defined at a future date. 85 phx20459 Volume Reloader [verified] reload_volume_ argument discrepancy. As time and resources permit this problem will be evaluated and better defined at a future date. 84 phx20315 Volume Retrieval [investigating] directory retrieval failed using -subtree when the pathname was one using an add_name for the directory. The add_name was what was supplied by the user.. As time and resources permit this problem will be evaluated and better defined at a future date. 83 phx20219 IO Daemons, Absentee Facility, Volume Retriever [verified] It appears as though the user can pass invalid information onto the daemons and absentee facility, causing either sub_err_ conditions to be raised or an output request's "Requested" time to be set to an arbit As time and resources permit this problem will be evaluated and better defined at a future date. 80 phx19037 Volume Dumper [investigating] lock wait time too short. As time and resources permit this problem will be evaluated and better defined at a future date. 79 phx18852 Volume Reloader [investigating] consolidated dump map not rebuilt on reload. As time and resources permit this problem will be evaluated and better defined at a future date. 78 phx18763 Volume Dumper [investigating] inconsistencies in volume data bases. As time and resources permit this problem will be evaluated and better defined at a future date. 76 phx18652 Volume Retriever [error] (volume_retriever) Use of the -subtree argument in volume retrievals generates proxy requests. A sufficiently large hierarchy blows out the volume_retriever queues.. As time and resources permit this problem will be evaluated and better defined at a future date. 75 phx18545 Volume Dumper [verified] The entry dmpr_report_$error_output doesn't create very good unique names for its error files.. As time and resources permit this problem will be evaluated and better defined at a future date. 74 phx18518 volume dumper and manager_volume_pool_ [error] The Volume Software nails your default volume segment.. As time and resources permit this problem will be evaluated and better defined at a future date. 73 phx18462 Volume Retriever [investigating] if volume retriever cannot recover an object after recovering the branch, he fails to clean up after himself.. As time and resources permit this problem will be evaluated and better defined at a future date. 72 phx18419 Volume Dumper [error] A typographical error in telling the volume dumper what tape to use results in the dump to be aborted. As time and resources permit this problem will be evaluated and better defined at a future date. 71 phx18417 Volume Dumper [verified] hc_dmpr_primitives should provide all available info when encountering an error so corrective action can be performed.. As time and resources permit this problem will be evaluated and better defined at a future date. 70 phx18357 Volume Retriever [investigating] In certain cases the volume retriever will attempt to search tapes that it cannot possibly satisfy a request from.. As time and resources permit this problem will be evaluated and better defined at a future date. 69 phx18049 Volume Backup Tools [error] The -test option to purge_volume_log fails to unlock pvologs under certain conditions.. As time and resources permit this problem will be evaluated and better defined at a future date. 68 phx17936 Volume Dumper [investigating] tape mount problems leave dead pvolog segments around.. As time and resources permit this problem will be evaluated and better defined at a future date. 67 phx17735 Volume Retriever [investigating] incorrect type in message for cross retrievals. As time and resources permit this problem will be evaluated and better defined at a future date. 66 phx17577 Volume Retriever [investigating] Will not volume retrieve links.. As time and resources permit this problem will be evaluated and better defined at a future date. 63 phx17165 Volume Retriever [investigating] "Failed to append branch for ^a because object already there with other name". As time and resources permit this problem will be evaluated and better defined at a future date. 62 phx17164 Volume Retriever [investigating] "Failed to delete user queue entry; Message not found.". As time and resources permit this problem will be evaluated and better defined at a future date. 61 phx17011 Volume Dumper [verified] retry strategy in dump_volume_ ineffectual. As time and resources permit this problem will be evaluated and better defined at a future date. 60 phx16964 Volume Retriever [verified] There is no facility to Volume Retrieve a level-1 directory.. As time and resources permit this problem will be evaluated and better defined at a future date. 55 phx16727 Volume Retriever [verified] Partial completion of volume retrieval can result in disconnect failures.. As time and resources permit this problem will be evaluated and better defined at a future date. 51 phx15590 Volume Retriever [limitation] Renaming a project generates retrieval problems.. As time and resources permit this problem will be evaluated and better defined at a future date. 48 phx15579 Volume Retriever [error] Linked TRs: phx13481 Retrieving a mailbox should not require a ring_1 process.. As time and resources permit this problem will be evaluated and better defined at a future date. 47 phx15458 Volume Retriever [investigating] Should adjust bitcounts to character, NOT word boundaries.. As time and resources permit this problem will be evaluated and better defined at a future date. 45 phx14444 Volume Retriever [verified] Some error messages have about 20 trailing blanks.. As time and resources permit this problem will be evaluated and better defined at a future date. 38 phx17732 A recovered volume log is not truncated or its bit count set properly. 37 phx16302 If the error message at line 151 of merge_volume_log.pl1 (null entry in both logs) is emitted, it will be repeated forever since neither i nor j are advancing within the loop. Also if the two logs being merged contain duplicate entries, then both entries are included in the merged log. Now this is not strictly within the mandate of merge_volume_log (which is supposed to merge a newly created volume log with an older copy) but there are instances where this arises. 36 phx14574 The module dump_volume_ attempts to handle and recover from a couple of error conditions during dumping -- segfault errors by retrying the output up to ten times, and page_fault_error by skipping the current object. The technique employed for both of these can result in serious problems if the tape thus created is subsequently used for a reload. The problems may potentially extend to many segments not directly involved in the original dump errors. When writing an object to tape, dump_volume_ calls dmpr_output_ to write both the backup header and the actual data object. The length of the data object is stored in the header in backup_volume_record.rec2_len. If a segfault or page_fault_error occurs during the output of the object, then the resulting tape_mult_ image for the object consists of a header which says that rec2_len characters have been written to tape for the object and an arbitrary number of characters (less than rec2_len) which have actually made it to the tape. When the reloader goes to read such a tape, the following occurs: it reads the header record which indicates that the object consists of the following rec2_len characters on tape. It issues a read for rec2_len characters and gobbles up both the partial dump of the object PLUS the header and contents of possible many other segments to satisfy the rec2_len specification. When it attempts to read the next header on the tape, it realizes that it is out of synchronization and attempts recovery. By now, however, the original segment which sustained the dump error contains bad pages consisting of whatever was on the tape to a length of rec2_len characters (including backup headers and pages of other user's segments); segments which were supposedly dumped without error after the original segment are not reloaded at all because they have been absorbed into the original segment up to its quota of rec2_len characters; and the segment encountered after rec2_len characters is also lost because of the reloader skipping during resynchronization. 30 The Volume Dumper must force write each database after modification. This could cause some degradation in dumping, but esd-less crashes would be less painful when one finds a database with null pages just when one wishes to use it in a hurry. 28 phx15580 phx17205 This is a generalized placeholder for Volume Retriever performance problems in regard to subtree retrievals. I am not convinced that there is much that can be done, but just in case... 26 phx16930 error in the implementation of -stop_vtocx 24 phx16692 Companion entry to 626. This holds the bug in the Volume retriever that mis-uses the authorization passed to hardcore to append a segment. 19 phx13032 retrieve_from_volume, the -list control argument causes all other control arguments to be ignored (silently). It also says nothing when there are no pending retreievals. 11 phx13031 If the uid of an object cannot be determined, ie the parent dir does not exist, the message "Online records indicate that X cannot be found ..." This is a truly mis-leading error message. 10 phx13032 The error message from retv -list -any_other_arguments is too terse. 8 phx11848 A window exists between closing one dump volume and opening the next (volume log entry) in which a restart will get an incorrect cycle_uid. This will make reload groups inconsistent. ----------------------------------------------------------- Historical Background This edition of the Multics software materials and documentation is provided and donated to Massachusetts Institute of Technology by Group BULL including BULL HN Information Systems Inc. as a contribution to computer science knowledge. This donation is made also to give evidence of the common contributions of Massachusetts Institute of Technology, Bell Laboratories, General Electric, Honeywell Information Systems Inc., Honeywell BULL Inc., Groupe BULL and BULL HN Information Systems Inc. to the development of this operating system. Multics development was initiated by Massachusetts Institute of Technology Project MAC (1963-1970), renamed the MIT Laboratory for Computer Science and Artificial Intelligence in the mid 1970s, under the leadership of Professor Fernando Jose Corbato. Users consider that Multics provided the best software architecture for managing computer hardware properly and for executing programs. Many subsequent operating systems incorporated Multics principles. Multics was distributed in 1975 to 2000 by Group Bull in Europe , and in the U.S. by Bull HN Information Systems Inc., as successor in interest by change in name only to Honeywell Bull Inc. and Honeywell Information Systems Inc. . ----------------------------------------------------------- Permission to use, copy, modify, and distribute these programs and their documentation for any purpose and without fee is hereby granted,provided that the below copyright notice and historical background appear in all copies and that both the copyright notice and historical background and this permission notice appear in supporting documentation, and that the names of MIT, HIS, BULL or BULL HN not be used in advertising or publicity pertaining to distribution of the programs without specific prior written permission. Copyright 1972 by Massachusetts Institute of Technology and Honeywell Information Systems Inc. Copyright 2006 by BULL HN Information Systems Inc. Copyright 2006 by Bull SAS All Rights Reserved