Jump to content

jg167

Members
  • Content count

    11
  • Joined

  • Last visited

Community Reputation

0 Neutral

About jg167

  • Rank
    Occasional Forum Poster
  1. My server running 6.0.204 on MacOS X 10.3.9 has been backing up 10.4 clients (and has resoted them as well) running client 6.0.109, but last nights backup aborted about half way into one client with the above inconsistency check. Is this known and fixed in one of the updates?
  2. I am happy to report that (as expected) loading a new 6.0 instance does preserve the config in the Retrospect preferences, but more importantly that 6.0.204 fixes (with just one little issue as seen below) the problem of a fully restored disk not booting. Here is what happened 1) restore entire disk to a newly initialzed disk (I happened to do this by booting the newly disked cube in target mode so it appeared as a FW drive on my main backup server just to be faster than their usual airport connection) 2) boot the new disk -> missing system folder icon 3) boot the new disk with option held down to see a list of boot possibilities and the new disk is indeed not shown. 4) boot from an external disk, run the startup disk sys pref and the new disk IS shown 5) select it and restart 6) all is well. So I'm not sure why the extra little two step was needed, but in the end it worked, so I continue to be a happy retrospect user (cube 1 as backup (and iTunes) server, backing up itself plus a 2nd cube, a Ti, and a half linux half XP PC).
  3. This is all relative to Mac OS X 10.3.x and the associated Retrospect files (ii.e. this is not on Tiger or using the Tiger Retrospect version). I am running 6.0.178 on a small network and am having the problem of not being able to boot a fully resotred disk (it hangs at the gray on gray apple logo screen with no spinning bar). The readme for the current download says that build 204 fixes a problem that sounds very much like this (i.e. a full restore of a bootable volume onto an erased disk where the restore includes many incremental MacOS X upgrades). However this download seems to be the full product and there is nothing in the readme that I find that says this will preserve the current configuration, licenses, scripts, etc. So is this safe to re-install or will I have to reenter all of my configuration? (I assume its the former, i.e. is safe, but thought I'd check first). The user guide talks about things to do when going from 4.0 -> 6.0 and from 5.1 -> 6.0 but I see nothing about just a newer version of 6.0 . -jrg
  4. jg167

    Linux Client No Longer Seen

    Well I give up. If I set the IP on my linux box manually retrospect can test the cleint and sees it fine, but can not find it when searching. But if I use dhcp to get an IP for hte linux box it works. It must be something i am not setting up manaully quite the same way, but the annoying part is that the retrospect "test" feature works, (i.e. this is not an effective test in this case as it passes when, according to the serach results, it shold not). Oh well, to get around the problmes of DHCP with a small server, I just set the least time to 10 days so its very unlikely the IPs will ever change.
  5. jg167

    Linux Client No Longer Seen

    I have a simialr issue that I have been unable to resolve. Mac OS X 10.3.3 (now actually 10.3.4 but this happed with 10.3.3 as well) backup server with Retrospect 6 client (one of 4) is redhat 9 All was fine until I changed the linux system to have a static IP instead of using DHCP. Now it is not seen by the backup server. I have uninstalled/reinstalled on linux, removed the old entry from the backup servers client db, still no joy. If I use Retrospects volumes/confugre/test feature and type in the IP it finds it fine yet does not when searching. It is in the same subnet. Suggestions?
  6. After playing with rules etc, I still think allowing a backup script to separate the backup specification (what, when) from the backup set so as to allow (e.g.) a 4 week cycle that uses one set, then the other (alternating every 4 weeks rather than alternating every week) would be better. Its less confusing and allows for file retention going back much further and contiguously. Bootable client CDs for all clients is still high on my list. Chained events would be nice (i.e. after this, do this). You can pretty much do this by just setting a sufficiently large time interval between them, but this invariable waits much longer than needed if it always waits long enough. Pre and post processing (apple, shell) scripts would be nice too. I'd like to see an option to remove the notice dialog and just put such things in the logs for the expected errors (i.e. in my config one or more clients is always not on the net, so this is not really worthy of a dialog message). Waking up sleeping clients is also needed. That being said, it IS working for me, so overall I'm a happy customer, I just could be happier!
  7. Quote: Quote: But Dantz has specifically excluded this card from its list of compatible SCSI host adapters. As such, system crashes when Retrospect attempts to control a device attached to the card would be expected. Dave Well as such I'd expect Retrospcet not to function, but there shold be nothing an application can do to crash the kernel. Since retrospect has already mentioned the problem I'd assume this has already been repoted to apple, but it definetly involves a kenrel bug of some sort.
  8. I agree, doesn't seem like that big of a deal to me either. They can snag all the info they need when the script is being setup and you first "log in" to the remote machine.
  9. thanks Nate. Right, that script is better than what I was envisioning, though maybe a little confusing when it comes to where to look for a restore. Being able to reference a group of backup sets by a virtual BS and having a scriptable action to move the virtual BS to a real BS name, or to the next BS in the group, might be an easier to understand UI. Which makes me think of something else. Suppose I rename a backup set (and its catalog) so when Retrospect comes up the backup set name is there, but now is a different file (still a backup set). e.g. if a and b are both backup sets, and I have been backing up to a, then do mv b tmp mv a b mv tmp a Is that giong to confuse retrospect when I continue to backup to a (which used to be or when I restore from either? Yes I can appreciate waking up a sleeping machine is not a simplistic thing. But (as many have said I'm sure) it would be really nice. Having bootable clients for all 3 system types also realy seems needed to me. Without this a full restore will require booting from an alternate source which is a hassle, though certainly easier on the Mac then XP or Linux. thanks again.
  10. After using Retrospect for several years on individual Macs I have just moved to Retrospect 6 and am using it to backup my home net consisting of the server (Mac 10.3.3) itself, plus 2 other Macs (10.3.3) and a PC (that runs both RedHat 9 Linux and XP/Pro). My background is system sw development (*nix, OS, dbms, even backup (NDMP) development :-). So far it seems to work ok, but I have some suggestions (some harder than others to implement). 1) Change the defaults to be backup all files EXCEPT caches (why you would EVER want to backup cache files is a mystery to me), and to backup only SELECTED volumes (on Linux there are almost always multiple volumes and almost always not all of them are in need of backup). The second one is especially misleading since you can select volumes, yet the backup is not restricted to only those unless you have the proper mode set. 2) Add more flexible backup set handling. I am using nearly the default setup of a normal backup every night for 4 weeks and then recycling. However this opens a window of vulnerability during the time the the recycle backup is being done (i.e. if a system crashes after the recycle starts but before it completes then no backup is available). What I want (and what is more common in my experience) is to simply alternate between two backup sets. So the recycle backup is always done to the OTHER set and then the 4 weeks of normal backups also go to that set. I think this IS possible now by having 2 scheduled events each 8 weeks apart for the recycles and 8 scheduled events for the daily normal backups (1 event per week) between then, but that is WAY too confusing to want to try out. The combination of normal and recycle backup also does not seem to be well coordinated, as it works by NOT doing a normal backup on the day that the recycle backup is done, thus this day is missed for the 4 weeks of normal backup. Thought I'd add for those interested, that I get around this, and add media redundancy by using rsync to a 2nd external drive identical to the backup target. So normal backup 3AM M-F to backup0, rsync every 10PM F backup0 to backup1, and recycle backup every 4th Sat 12AM to backup0. That way if something goes wrong with the recycle I have a week to notice it and use the rsync copy. 3) Make it wake up sleeping machines. This would seem possible by issuing a harmless network administrator action and setting the Mac to wake on such action prior to attempting access. Not sure if sleeping linux and XP are a problem yet. 4) Produce bootable client application CDs for all machines. If a system needs to be restored and is not bootable, if its a Mac I can (I think) use the bootable Mac CD to restore it from the server (i.e. running the client after booting off the CD), but I see no such feature for XP or Linux (without buying the server for XP that is). Bootable client CDs are a must. 5) Allow a single mutli-boot system to use one license. e.g. My PC runs both linux and XP, I'm happy to only have the one that happens to be up at the backup time to get processed in that backup but the extra $50 seemed a bit excessive. -jrg
×