July 26, 201114 yr A hot spare disk that sits idle and will automatically be reassigned to replace a failed drive (not the warm-spare cache drive method). Useful when there are 2tb drives in 16 plus drive arrays.
July 27, 201114 yr This has been suggested before, but ultimately the consensus is that it is a bad idea. There are a lot of reasons that a disk can fail, and not all of them are due to problems with the disk itself. The issue could be cabling, insufficient power, motherboard issues, etc. When a disk fails, it is important to evaluate the situation and carefully choose the correct action. In some cases replacing the drive is the correct action, but not in all cases. A drive that automatically steps in to replace a failed disk can cause even more problems in certain cases. This is one of those instances in which no amount of technological innovation is going to replace the flexibility of human judgement.
July 27, 201114 yr With current unRAIDs you can have a hot spare, but you must manually tell the array to start using it - just stop the array unassign the disk you want the spare to replace, assign the spare drive into the array and restart unRAID (no need to do a full power down since the spare drive is already powered up and atached to SATA). If the spare has already been precleared the rebuilding will start pretty quickly. I've done this a few times when upgrading disks, I preclear the new disk first and let it sit (powered up and connected to SATA), then once I've done a parity check and a backup I will do the manual swap. Regards, Stephen
July 28, 201114 yr With current unRAIDs you can have a hot spare, but you must manually tell the array to start using it - just stop the array unassign the disk you want the spare to replace, assign the spare drive into the array and restart unRAID (no need to do a full power down since the spare drive is already powered up and atached to SATA). If the spare has already been precleared the rebuilding will start pretty quickly. I've done this a few times when upgrading disks, I preclear the new disk first and let it sit (powered up and connected to SATA), then once I've done a parity check and a backup I will do the manual swap. Regards, Stephen When replacing an existing drive, the preclear signature is not used at all. unRAID always writes the entire replacement drive. It will always start quickly. The pre-clear signature will prevent a lengthy server down-time when adding an additional drive to an existing parity protected array. That is not the same as replacing a failed drive. It is good you pre-cleared it, and have some level of confidence in it, as about 1 out of 5 drives lately seems to have some kind of problem when new.
October 5, 201114 yr Old thread, but I wanted to add my +1 for this request. I get why it's best to have a human look at the problem.. but that's not always possible. For those of you who are home looking at your unRaid boxes every day - great. For those of us who aren't home (I work at sea), this is a great option to keep the array protected until I get home a look at it. This shouldn't be hard feature to write.. just add a selection for a hot-spare to the array setup screen (as you would select a cache drive), and then just fail over to it. A nice big warning to prevent people who don't need it from using it, and all should be well. This is a common feature in larger commercial solutions, so why shouldn't it be looked at for unRaid? Cheers, whiteatom
October 5, 201114 yr +1. I'm Pro Optional hot spare also I don't think it's a bad idea as previously mentioned in this thread. I would agree that any drive kicked out of the array needs to be investigated for reasons. It may have been kicked out without the need for rebuild. Yet I would prefer the option of a self healing array vs having to stop the array and assign the drive manually.
October 5, 201114 yr I also would like the hot spare option. If you're away from the server and a drive fails for any reason, if unRAID had a setting to automatically remove the failed drive and start using your hot spare, I see how no harm can come of this. Sure, sometimes the failed drive didn't really fail, and that's great. Fix whatever issue it had and put it back in the array, or clear it and use it for your hot spare now.
October 5, 201114 yr +1 While it is true drives fail for many reasons, replacing that failed drive is probably a good idea. Hot spare and auto rebuild will enhance data protection. Motherboard, cable, or power supply problems will not be resolved but those don't place data at risk.
October 6, 201114 yr If anything I would like a simple click click enable hot spare into array function opposed to an automatic function. Currently I have a drive pre cleared sitting in a box in case there is a needed swap out.
October 6, 201114 yr If you look back in the forums at people that have had drives get the red ball, you would see that the reason is seldom an actual failed disk. More often it is a bad or loose SATA cable, power splitter, backplane, or controller. Depending on what it is, automatic rebuilding could make matters worse. (e.g., syslog could consume all memory and crash the server). If the problem happened to be heat related (maybe your HVAC or fan went out and a drive failed due to the heat), the worst thing to do is spin every disk. I have no objection to the feature but do believe users would be better served letting unRaid simulate the disk until the human can make the call on next steps.
October 7, 201114 yr It is also rather common for a second drive to drop out of an array if you try to do a rebuild when you are having a cabling or other problem not caused by an actual bad disk. It is much more difficult to recover from an array that has two disks drop than one.
October 11, 201114 yr Is it not preferable to have a dual parity option and be able to support 2 failed drives? I will prefer this option as it will make the array safer and don't automate any recovery process that implicates writing on a disk. The system can be inteligent also and let me wirte in the array with one failed drive, but only read with 2 failed drives or something like that... Well, I will see you opinions on that.
October 11, 201114 yr Does the 5.0 series allow an exernal program or script the ability to swap/assign disks? Could some type of plug-in be written that would stop the array and swap a disk be written for the newest 5.0 beta series? Peter
October 12, 201114 yr The system can be inteligent also and let me wirte in the array with one failed drive, but only read with 2 failed drives or something like that... It would not be difficult to manually mount unRAID disks in read only mode, even if the array won't start. Does the 5.0 series allow an exernal program or script the ability to swap/assign disks? Could some type of plug-in be written that would stop the array and swap a disk be written for the newest 5.0 beta series? I don't know of a way to automate the assignment of disks to an array - unless you are directly manipulating the super.dat and/or disk.cfg files. Best you could do is some sort of remote control of the unRAID GUI.
June 21, 201214 yr Any work on this from limetech? I have seen in other threads that Tom has said he's interested - but no comments from him. whiteatom
July 5, 201213 yr Any work on this from limetech? I have seen in other threads that Tom has said he's interested - but no comments from him. You likely won't hear anything on this until after v5 goes final, that will be his primary focus at the moment as it is well overdue. It will be worth the wait though, anyone using the beta can tell you that. Whether or not hot spare gets implimented as an option will be up to Tom (or possibly a 3rd party plugin) but dual parity is much more likely, though not in the near future as AFAIK this requires almost a complete re-write.. v6..? perhaps
Archived
This topic is now archived and is closed to further replies.