Jump to content


  • Posts

  • Joined

  • Last visited

  • Days Won


Everything posted by denywinarto

  1. Topic, my drivepool is also pending update. Any idea why?
  2. Thanks for sharing your experience, thats also what i've been using together with stablebit scanner. My only concern is that it lacks daily scanning scanner has, or at least i havent figured out how to do it since i'm used to rely to scanner. Sentinel also has this estimated lifetime, but i found out it wasn't always accurate as well. One of my 6tb red wd died before the estimated time. But then again my experience with scanner wasn't much better, iirc it only warned me a smart error less than 24h before the drive died. For me now whenever a drive shows filesystem error multiple times im replacing it right away. Edit : i forgot to add, some of my drives were failing even at 100%, so its not a guarantee.
  3. Just windows defender. I have ticked all options and i'm pretty sure that one as well. You're right about SS it could be really a PITA. But i dont see any other options at this point that could stripe 4x3tb drives. Drivepool doesn't seem to capable. Even 8TB ssd still costs $800, 3x 4tb wd black is not even half the price, not to mention i have to sell these at less price. I suspect i will have to rebuild the Storage space pool from scratch, makes sense somehow because these drives have been migrated from a different motherboard, (but it was working fine for 3 months after migration) I haven't actually tested storage spaces running without Sb scanner on my main server, just on my spare server with exact same motherboard and OS (it was fine for 3 days with scanner service stopped) It's too risky since i'm definitely sure it messes up the OS. Just bought another HDD for the backup, i'l ltry recreating storage spaces from scratch once it arrives.
  4. Bad news, this is what i did. Made a backup Installed 3x4 TB wd in the rack, (In my other spare machine with same motherboard and cloned OS it had no issue) The SS Disks were offline. So i thought i'd exclude it from scanner's surface test lists first. Install stablebit scanner, exclude it from scanner It does appear on scanner as "Microsoft storage space drive" IIRC. All of a sudden BSOD, haven't even re-initiate the disks yet. So i unplug the 3x4 TB. ill got BSOD. So i restored the backup before scanner installation, thankfully it appears to be stable until now. So conclusion is, excluding it from scanner's list doesn't work.. I'm gonna check the SS drives on my spare machine again, but last time it was repaired and there were no issues
  5. Okay, i'm not expecting a fix in 3 days or something like that, Just a word acknowledging this issue and a confirmation of future version with fix would be nice. In my opinion this should be regarded as critical issue because its affecting OS stability. About storage spaces, Well, I need 12tb with ssd-like read speed, with the current price of SSD i can't find any other alternative which offers similar performance to size ratio, and i'd rather stick with what i have now. Maybe once SSD price goes down i'll reconsider it. I plan to make OS backup and run scanner and turn the scanner off for the storage spaces drives. If anything goes wrong i can just revert to the backup, but hopefully i wont need to do that.
  6. Did you read my post about the chronological order of BSOD? windows server 2019 + corrupted Storage spaces MBR= no BSOD, windows simply gives warning "this drive needs to be scanned blablabla" windows server 2019 + corrupted Storage spaces MBR + stablebit scanner running = BSOD, right after inserting license code and scanner starts running. As you can see on the reliability pic on my last post. The fix is for scanner is quite simple, the trigger is bad MBR sector from Storage spaces harddisk, so simply prevent it from scanning once it detects corrupted SS drive. I know this is stablebit forum, but without customer critics a product won't get any better. the 3 drives are now fine in SS after chkdsk, without stablebit scanner of course.
  7. 3 days and not a single response? Not even my ticket gets responded Scanner is an abandoned product then, Anyone know a decent alternative?
  8. Update : thankfully CHKDSK saves the day. it says the MBR was corrupted but it was able to repair it So now i can probably use my storage spaces again But should i use scanner or not now? Any words from Chris / Alex or the devs? Can't risk another BSOD
  9. Well got another case now with stablebit scanner. This time its worse, its 3x4 tb wd black in storage spaces. It became RAW after BSOD I'm pretty sure its scanner cause as soon as it starts scanning after installation BSOD occurs. And the reliability windows page back this up. (I installed scanner on 6.36 and input the serial about half an hour later.) Its not drivepool cause the drives arent in the pool. This is on newly installed Server 2019, i just reinstalled it yesterday I immediately uninstall scanner. What should i do now? I need scanner to warn me of bad drives so i can replace them before they go bad. But i have 2 dozens of drives if a single bad drive + scanner = BSOD then that means 2 dozens reliability holes. Edit : I should have add, The reason i reinstalled my machine was because i'm having multiple BSOD's everyday. Now i suspect my recent midnight BSOD's were caused by scanner scanning at midnight. 1. A new installation BSOD'ed right after i install scanner and it's scaning 2. The BSOD always happens at midnight, i check task scheduler and nothing is run at midnight. Seeing scanner's description of "always scans at midnight" reminds me of this. While i know the bad drives are the fault, i think scanner could do a better job preventing windows BSOD trigger
  10. You dont need to disable any service if that's what you mean
  11. 1. Manually move all files to the outside of the poolpart.xxx folder on the bad disk so that drivepool can't see them any more. 2. Remove the old disk from DrivePool (it will be instantly complete since the poolpart folder is empty). 3. Insert new disk (new different PoolPart.xxx is created). 4. Manually copy the files from the old disk to the new poolpart folder on the new disk. I found out by doing this i didn't get BSOD compared to moving the disk using DP Above is why i think the BSOD is related to drivepool, what i'm worried about if one my drives go bad then it would just throw random BSOD and screw up my OS
  12. So you're saying its more like OS issue? Hmm i'm using server 2019 and LSI 9300 8e.. Any possible workaround that is recommended? I mean eventually all my drives are gonna go bad someday, just wondering if there's a way to prevent it from screwing up my OS installation
  13. About 2 days ago i instaled 1tb hdd which i thought was still in good condition because HDD sentinel and stablebit scanner didnt report anything strange. Last night i think my server got rebooted couple of times and some drivers got corrupted. Thankfully i made daily backup, so i restored it. But then just few hours ago it rebooted again, and that 1tb HDD became "RAW" and offline. I also remember when moving bad drive from the pool last year i got BSOD once or twice. So Is this a known issue? I'm using the latest build.
  14. Any links for this? Have you seen people achieve it? I could make windows detect them both in device manager, by still enabling the svga, but running gpu-z gives me svga as the detected gpu. I'm also using the igpu for blueiris so im forced to use windows
  15. Yeah, strangely the machine script says my lsi-9300 8e cant be passthrough'ed So I tested with esxi as well. It worked but there is one issue that makes the setup not ideal, to passthrough the igpu, i have to disable svga, thus i lose console display and i have to rely on rdesktop alone. Not ideal for 24/ server IMO Are you by any chance also pasthru your gpu / igpu?
  16. Ah i thought you were using hyper-v lol.. Thanks for the license tips will keep that in mind. I'm leaning towards hyper-v cause most of my environment is windows based..
  17. Thanks chris, how long have you been doing it? Is it stable enough? Also there is something i forgot to ask, were you able to passhthrough the igpu as well? I will need it for emby and blue iris i think
  18. Hi Philmatic, sorry for bumping an old thread, I thought controller passthrough is possible already on server 2019? https://www.reddit.com/r/HyperV/comments/bc3vvc/pcie_passthrough_possible_with_a_hba/ Have you tried it? If it does passthrough the SMART status as well then it's probably the better method
  19. Anyone ? Chris? Have you seen this implemented without issue?
  20. I'm in a process of migrating to hyper-v 2019, been reading some forum post but still a bit unclear about the process. 1. Should i passthrough the drives or the HBA card? With DDA looks like it's possible to just passthrough the HBA? 2. I read SMART isnt passed if i passthrough the drives, is this also the case for HBA passthrough?
  21. A blackout just occured and upon startup stablebit scanner gave me this error One or more disks are expected to fail within 24 hours The detail says there are 6 unstable sectors iirc.. But about 15 min later that error vanished, i couldn't find it again. The same drive showing healthy now.. I checked with hdd sentinel and it says that drive is healthy So is this some kinda false positive cause its the first time i got this. Should i replace the drive?
  22. Sorry forgot to update, i managed to recover my data. and yes, drivepool had to be restarted. Painful (and expensive) lesson, but it taught me one thing, never ever create RDP shortcuts to your important servers
  23. This morning my server was hacked and the hacker encrypt all of my drives with bitlocker. Now he gave me a password to unlock it and it was successful so i unlock all drives. But now on bottom right tray there's a multiple encrypting drive X... in progress.. There are over 180 TB of data. Now what i dont understand, drivepool still can't detect the drives. I am able to open the files on the unlocked drives though, do i have to wait until encryption process completed? what i dont understand is, isnt' it supposed to be decrypting instead of encrypting?
  24. Thought i'd answer this here since my new server's been running stable for almost a month since the issue. The root of the issue was duplicated pool, as evidenced with whocrashed report showing drivepool's filesystem as one of the trigger. So the lesson is as i mentioned before, never try to create duplicated pool. I also learned that mounting hdd to folder is not really necessary if you use drivepool entirely, kinda makes new installation easier since we dont have to re-mount the drives
  25. Specs : MSI Z390 Pro Intel i7 8700 32gb ram corsair HGST 4u60 LSI 9300 8i Intel i2Ti Last night there was a power surge, not sure what caused it, and it fried my nvme i think, it became undetected So i reinstalled a new nvme drive with server 2019. During installation it went just fine. But when i try to initialize drive, it freezes. And Got this BSOD screen saying Unexpected kernel mode trap What failed C1.dll The thing that bothers me is that it has been working fine for almost a year with server 2016. What i have tried 1. Full install all driver > resulting in spinning logo for 1 hour, so i just installed the audio and intel iris driver 2. Change the LSI 9300 8i, i happen to have a backup, but it's still the same thing, so its not sas card issue 3. Change the memory, 4. Remove the intel PCIE nic 5. Move the SAS card to another PCIE 6. Removing the cable plugged to the SAS card > this allows the windows to load correctly! But i cant access the jbod I was also wondering if corrupted drive or unfinished drivepool can cause this? Because on my very first attempt to reinstall, everything actually worked, but i installed an old drive with ntfs mounting in it. And drivepool created 2 pool, i was afraid it might mess my library, so i stopped DP service and i reinstall again, but now im stuck The only thing left is the motherboard and the rack... And i happen to not have any backup for them.. Been struggling for hours now, any idea? Edit : i swapped the mobo to lower end spec'ed and it seems to be stable now Now interestingly, at first i got 2 BSOD, during this time theres a duplicated pool that i mentioned earlier https://imgur.com/gallery/QPPOjcM But then after i turn off the drive from duplicated pool, delete poolpart, and it went back to 1 pool. It seems to be stable, no more crash until now, about 5 hours So its either the motherboard or incorrect drivepool pooling... Edit 2 : im gonna put this here, this is how i solved duplicated pool and how it was created : 1. Turn off the drive from the pool that i don't want 2. stop DP service 3. Turn on the drive from the pool that i don't want 4. delete poolpart.xxx from that drive 5. start DP service 6. Duplicated pool disappeared Also in my case, i had the old drive with previously NTFS mounted folders installed as secondary drive, cause i want to copy the old content (emby, radarr, sonarr, etc) while i want the drives to mount to the new NTFS mounted folders i created on the newly installed drive And this was a BAD idea, The initialized drive went straight mounted to the old drive, even after i delete the NTFS mounted folders to recycle bin. and this is how i got the pool initialization mixed up So lesson learned i guess, i will never use 2 drives if the older one has older NTFS mount folders
  • Create New...