greybeard

Members
  • Posts

    127
  • Joined

  • Last visited

Everything posted by greybeard

  1. I have a 19 dirve 6.6.6 array. 1 x 8 TB Parity 18 x 2 TB MBR unalligned ReiserFS data drives. 1 empty drive bay. I also have 4 x new 8 TB drives so what I want to get to is 2 x 8 TB Parity 3 x 8 TB GPT alligned XFS data drives. 8 x 2 TB XFS data drives. Don't think it makes any difference if these are GPT or MBR. Pretty sure I can not replace a 2 TB ReiserFS unalligned data drive with an 8 TB XFS allinged drive just by doing a disk swap and rebuild. So here is my idea for how to get converted. Create a 3 x 8 TB array on a spare box. From Win10 system, copy date from 11 of the 2 TB drives to the 3, 8 TB drive array. (The 2 TB drives are all 98% full. I know that is not best practice.) From Win10 system, verify copied data. (I have a way to do this.) Shut down both arrays. Create a new config with Replaced parity drive with a new 8 TB drive. Replaced 3 of the 2 TB drives with the 3, 8 TB drives I copied the data to. Remove the other 8, 2 TB drives, i copied the data off of, from the config. Start the array and build parity. If anything goes wrong with 5 or 6 I am thinking I can restore the original config and original drives and have good parity. Wait for parity sync to complete. Preclear the 8 TB drives I coped the data from and are no longer in the arrary. (Technically I may not need to preclear but planning to do it anyway.) Stop the array and add the precleared drives. Start the array. Copy data from remaining 7, 2 TB drives to the now empty 2 TB drives. Verify copied data. Shutdown the array. New config. Replace the new parity drive with the old one. Remove the 7 old 2 TB drives from the array. Start the array and build parity. Again I should have good parity if I have to revert to the config i have at the end of step 15. Wait for parity sync to complete. Shut down the array. Add second 8 TB parity drive. Start array and let it build parity on the second parity drive. Do what ever I want with the 7 old 2 TB drives. Maybe preclear and add some back to the arrary to have more free space. Will probably take a month or so to complete all steps but when done I should be able to replce any of the 11, 2 TB drives still in the array with an 8 TB drive and get an XFS alligned partion after rebuild. Thoughts? Pitfalls? Other ideas? Thanks for taking a look.
  2. Like most I voted for 16b. However I think there should be consideration given to staying with the reiserfs patch, even for 5.0 final. The patch has been very well tested. Changing to the official fix would require rigorous testing and, who knows, might not even work. So for 5.0 I would suggest staying with what has been tested and works.
  3. The Upgrade Instrucitons for release 5.0-rc8a and older do not say anything about copying the syslinux.cfg file to the flash drive. Am I correct to asusme that is only because the instrucitons have not been updated and I really do need to copy it?
  4. Keep in mind that the code is for $15 off an order not a per drive discount.
  5. Using version: 5.0-rc6-r8168-test with C2SEE I was getting inconsistent response times when clicking on things in the web interface. Worked fine for a few minutes then got a 15, 20 second or longer wait before a response. I finally got tired of all the problems with the built in nic and decided to follow the advise so often given in these forums so I purchase this. http://www.ebay.com/itm/Intel-PWLA8391GT-1000Mbps-Desktop-PCI-Network-Card-NIC-/150592466899 Installed it and now my web page response is 100% consistent. After a few days of testing I have not once had any delay in web page displays. I suppose I will need to add one to my two primary systems before converting them to 5. Only mistake I made was not moving the on board jumper to disable the built in nic. Apparently unRAID uses the first nic it sees so it kept using the realtek instead of intel. After cracking open the case and moving the jumper all was good. On the C2SEE it does take moving a mobo jumper, can't disable it from the BIOS settings.
  6. There have been quite a few posts over in the 5 beta threads stating there is a real need for NFS and AFP. Maybe I can understand AFP for Apple users but I am more puzzled why there is such an urgent need for NFS. I have been running 4.7 with 20x2TB drives using SMB for a long time. Never have a problem finding the shares from my Windows machines, get 100m read speads, typically 35-40 write (no cache drive). So why do I need NFS?
  7. I used a Logitech Cordless MediaBoard Pro for PS3 in an HTPC I gave to one of my kids for Christmas. They tell me it is working great, when they need to use a keyboard. Two things. 1. It does not come with a bluetooth receive. You need to get one somewhere else. Extra $5. 2. Can't use it to get into the BIOS and make changes. Need to hook up a wired keyboard to do that. Also gave them a Media Center IR remote that they use most of the time. Media Center is started at boot up, and they rarely leave it. TMT5 for Blu-ray and MyMovies for catalog/menu. I know it says it is for a PS3 but works great with PC. Lot's of feedback at the egg, most seem to be using with a PC. It is a full size KB with track pad. All the keyboards I researched that have a joy stick or track ball got considerable worse feedback.
  8. Do you have any 2 TB Samsung F4 HD204UI drives in your array? If so, has the firmware been updated? Could be completely unrelated but is something to check.
  9. Not sure I would say that it is recommended. From what I have read it is either compatible with your main board and works fine or it's not. There is no inbetween. The evidence seems to suggest that if there is an JMB controller on the main borad you chances of this card working drops dramatically. So, unless you can find a post that confirms it works with your main board you spend your $9 (price went up) and take your chances. I have several of them working just fine in both unRAID and Windows systems.
  10. More goodness. I have a new build with three Hitachi 3TB cool spin drives. One on a mobo port, one on an MV8 and one a JMB card. Newly configured flash drive. First started testing with beta 9 and had some of the problems report in these threads. Just updated to beta 12 and everything I have tried (not that I tried a lot) is working perfectly. Web interface is snappy (wasn't in early tests) and a parity check started running at 117MB per second. Of course that will drop as it gets further into the drive. So, for now, things are looking really good. I will do some more experimenting and post the results. 15 bay box with C2SEE - E3400 - Kingston KVR1333D3N9K2/4G DDR3-1333 SASLP-MV8 and JMB PCI Express Hong Kong special. CORSAIR CMPSU-750TX I really hope that this means it is getting close to the end of the beta cycle. A stable 5.0 official release would be a great thing. I too have to thank Tom for all his hard work and everyone that has been testing these betas.
  11. Looks like Supermicro is updating the AOC-SASLP-MV8. I don't see it listed yet at the sites I normally buy from so I don't know what it is going to cost. Key Features • 8-channel SAS/SATA adapter with 600MB/s per channel • PCI-E x8 interface • JBOD mode only. No RAID support • Dimension: 6.6" (length) x 2.7" (width) More details here: http://www.supermicro.com/products/accessories/addon/AOC-SAS2LP-MV8.cfm
  12. I have been experiencing this behavior for a long time, have read other posts describing similar behavior and finally learned enough to have an idea what might be causing it. First, I am assuming unRAID is using ReiserFS v3. I have not learned enough to know how to check that but have read it is v3.6. If it is not v3 then the rest of what I am going to write is kind of useless. Ok, the problem: When a disk is getting close to full, somewhere above 90%, copying a file to it will often cause an initial pause of 20 seconds or so during which time the light on the data drive flashes continously. Only the data data drive, not the parity drive. Some times Windows will get tired of waiting and time out, resulting in an error. After the initial pause, the copy works fine, unless... There is other activity on the system, for example a preclear. In which case a block of data will be copied after the pause (parity drive light flashes for the actual data copy) followed by another pause, flashing data drive light, of the same duration as the first one. This repeats and repeats until the file is finally copied. A 500m (yes meg) file has taken an hour to copy in this situation. My theory: Based on research and observation I have come to the conclusing that what could very likely be happening is free space bitmap searching. When nothing else is running the bitmap is cached so there is only one pause, however if there is other activity, the cache gets purged resulting in another read of all the bitmaps spread across the partition. My question: Again being a UNIX newb I am wondering if someone can tell me if the patch described by this post is part of the unRAID distribution? It seems to describe exactly what I am experiencing although I never checked the CPU utilization (that newb thing again). Thanks for any feedback. Jerry http://www.mail-archive.com/[email protected]/msg21592.html
  13. Interresting subject. I signed up for a 2G account to checkout SpiderOak and kind of like it. However I would suggest that security on the share funciton is not nearly as good as it could be. I posted the following to the SpiderOak features request forum and am repeating it here, instead of providing a link, because their forums don't seem to allow public access. I think you have to have a SpiderOak account just to view. Here is what I wrote. I setup an experimental share to get experience with how it works. In the case where I want a share to be truly public, giving out the URL is a nice solution. However if I want to keep a share semi private, where only friends a family can see it, there is a risk. I can tell friends what the Share ID and Room Key is along with instructions to go to spideroak.com and use the share login but the result is the same URL as if I had sent them a URL. Now if all my friends and family were computer savvy folks and understood that the URL need to be protected to keep unwanted people out, that might not be a problem. My fear is that after logging in the URL will remain in browser history or be unwittingly published in a public forum thus exposing my data to the whole world. It would be a great security feature if there were some way for me to configure a higher security share that always forced family and friends to use the Share Login process. I would expect that clicking on a saved or historical link to the share would display a Share Login screen. I completely understand why you would not want all shares to work this way but it would be a really nice option. I also realize that my friends might give or post the Share ID and Room Key other people but that would be deliberate vs. accidentally disclosing the URL. And giving out the Share ID and Room Key would not expose the share content to indexing by search engines. The most secure option would be to allow a share to only be accessible to a specific list of other registered SpiderOak customers. Combine that with a feature to let those customers automatically download updates (kind of like a read only one way sync) to their system and SpiderOak would be an even better product. An option to restrict share access only to other customers would, in my mind, also be an acceptable alternative to always forcing a Share Login as I described above and could be a way to get more people to signup.
  14. I had a similar problem. 30+ days, no cards. I finally found them sitting at the post office waiting for us to pick them up. The USPS claimed to have tried to deliver them but did not leave anything telling us that. For four cards a signature was required so they wouldn't just leave them in the mail box. This sort of thing happens all too oftern with our local post office. It is incredibly anoying. The seller did offer to refund before I found them but I told them to wait until I did some more checking.
  15. Hope this helps. If you can find a flash chip, you have better eyes than I. Also scanned at 1600 dpi but the file is 21M. The back has a few more capacitors and resistors but that is it. From the 1600 dpi scan I was able to confirm it is in fact a JMB362 chip. I believe the ROM is built into the JMB362.
  16. At the risk of repeating myself: From the specs "Supports up to 64K Byte external Flash for BIOS expansion" Not on this card. Don't believe there is any way to flash this specific implementation.
  17. I have the $19 Newegg special C2SEE. Here is a clip from log showing the CPU. Mar 7 10:33:41 MiniClops kernel: CPU1: Intel® Celeron® CPU E3400 @ 2.60GHz stepping 0a (System)Mar 7 10:33:41 MiniClops kernel: checking TSC synchronization [CPU#0 -> CPU#1]: passed.Mar 7 10:33:41 MiniClops kernel: Brought up 2 CPUsMar 7 10:33:41 MiniClops kernel: Total of 2 processors activated (10374.33 BogoMIPS). (System) Hope that helps.
  18. Agreed. Should but for some reason didn't. It was like running memory at .05 volts too low. Sometimes it worked for hours and sometimes had problems immediatly and the symptoms would change. To be on the safe side I will reapply Stabilant to the drive edge connectors when I swap drives. The 5ml I have (thined out to 20+ml) will last many years and only takes a few seconds to brush on, that is if I can find it a year from now. It was much cheaper than replacing the M35Ts which is where the problem seemed to be. Only time will tell, few things in life are guaranteed. When I think about what actually goes on inside computers I am amazed that they ever work.
  19. Each preclear ran in it's own "screen" session. I had never read there was a limit but was wondering if it was going to work as I was starting them. Apparently it does work.
  20. Recently I finished building my second unRAID server. Centurion 590 Corsair 750TX Supermicro M35T bays C2SEE, E3400, KVR1333D3N9K2/4G SASLP-MV8 & JMB362 Variety of 1.5TB and 1TB drives, 15 total. When I first put this together and started testing I was having a variety of drive problems. Reported errors, slow transfers, drive not detected, etc. Isolating the source to a single component (cable, drive, bay backplane, power supply) didn't work. The problems would move from one drive to another, reseating a drive in the drive bay would change the symptoms or make them go away. Moving a drive to another bay would start out working only to see a slowdown popup later. In one test I created a 2 drive array and did the initial parity sync. Then started a parity verify, the verify speed slowed all the way down to 1.4, without ever logging any indication of an error. After a while I thought maybe I did not have good connections somewhere between the drives and mobo. Most likely where the drive plugged into the M35T back plane. I remembering having read an article in Byte magazine about something that helps makes electrical connections more reliable so went searching on the net. Quickly found something called Stabilant 22 and immediatly recognized the name. You apply it to edge connectors before pluggin in cards, cables or drives and it claims to make the connection much better. Maybe it sounds like snake oil but I found a lot of articles and postings saying it works. So I orderd a 5ml kit (not cheap $38 + shipping) and bottle of Isopropyl alcohol (another $8 with shipping) to thin it for applicaiton. When it arrived I took everything appart and applied to the card edge connectors, cable connection points and hard drive edge connectors as I reassembled the system. Did not touch the memory. It only takes a tiny amount (a fraction of a drop) for each connection so 5ml goes a long way. I don't think I even used 5% (about $2 worth) of it. Since I started the system back up I have not had one single error or slow down. I have run 15, 2 pass preclears, all at the same time, created a 15 drive array (9 drives had data, 6 new drives), did a parity calc and verify and ran CRC checks against one of the data drives. All went perfect. While this solved my problem the standad caution always applies. YMMV. Over the years I have built 30+ systems and until now have not had to resort to this. I have learned that most computer problems are bad connections. Most of the time just reseating fixes the problem but not in this case. Jerry
  21. Doh!! You are absolutely correct. Ultimately the 1.5TB will be the parity for an array of 1.5TB and 1TB drives. The 2TB in the box now will be moved to another box of all 2TBs. Guess I will need to use one of the 2TBs as parity for this test.
  22. Just finished testing using preclear. C2SEE, E3400, KVR1333D3N9K2/4G, CMPSU-750TX, Samsung 1.5TB HD154UI Test 1: Sil3132 RAID BIOS in slot 7 connected to Samsug 1.5 with 14 additional preclers running. 2 passes on all preclears. Results: Took 80 hours for all preclears to finish. The Samsun connected to the Sil reported that the preclear did "NOT" finish successfully and listed two skip/count messages. Test 2: JMB362 in slot 6 (primary video slot) connected to same Samsung 1.5 drive as Sil in test 1. Results: 21:24:42 Successful Test 3: Sil3132 RAID BIOS in slot 7 same 1.5 drive. Results: 21:45:34 "NOT" successful, one sikip/count message. Way different location than Test 1. Test 4: JMB362 in slot 7 same 1.5 drive. Results: 21:24:13 Successful Currently working on testing a 13 drive (5 1TB, 7 2TB, 1.5TB parity) array with the parity drive connected to the JBM362.
  23. The one I have is working great but I have not had it that long. Ran a full preclear of 1.5TB drive with no issues at all. Currently running the exact same preclear with the drive connected to a Sil3132. Seems to be going about the same. I have not tested with two drives connected to one card. I am kind of thinking I will connect my parity drive to this with data drives connected to a MV8 and mobo ports. Will be testing a lot more in the next few days with the goal of going live this weekend. Going to split a 20 drive server into 10-10 so all my 1 & 1.5TB drives will be in one server and 2TB in the other. Then I will add some more 1 & 2TB drives to the respective servers. In the end I will wind up with two 15 drive servers and spare capicity. Waiting for shipping is kind of a pita but they do have a lot of good feedback. I will probably buy a few spares to put on the shelf.
  24. From the specs "Supports up to 64K Byte external Flash for BIOS expansion" Not on this card. Don't believe there is any way to flash this specific implementation.
  25. The card arrived and it is working great. Even works in the primary video slot of a C2SEE. All of the SIL3132 cards I tried in that slot caused problems. Currently running a preclear on a Samsung 1.5 drive. at 5% complete into the preread I am getting 95.7 MB/s. About the card itself. There are no BIOS messages displayed during boot so I don't know if there is a way to turn of INT13, assuming it is even on in the first place. It presents as two AHCI ports both in unRAID and Windows 7. So in Windows there is no need to install a special driver. Windows loads the exact same drive as it does for mobo AHCI ports. I will preclear the same drive with the card in both an x16 and x1 slot then report the times. Will also provide the time for the same drive connected to SIL3132 in an x1 slot. I'll report back when they are done. I really like the way this is working and the JMB363 chip is listed in the Hitachi paper as being compatible with drives greate than 2.2TB. Not to mention at $8 you can't bet the price. Only downside is waiting two weeks for delivery, but I suppose that all depends on where you live. Native 4k sector support? That might be wishful thinking. No way to know at this time. The card itself is really tiny, too bad someone doesn't put several of these chips all on one card, each using their own PCI Express lane. The "RAID" in the description requires port multipliers to be connected to the card, no native RAID support like the SIL RAID BIOS. Feb 1 07:20:02 MiniClops kernel: ahci 0000:01:00.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16 (Drive related) Feb 1 07:20:02 MiniClops kernel: ahci 0000:01:00.0: AHCI 0001.0000 32 slots 2 ports 3 Gbps 0x3 impl SATA mode (Drive related) Feb 1 07:20:02 MiniClops kernel: ahci 0000:01:00.0: flags: 64bit ncq pm led clo pmp pio slum part (Drive related) Feb 1 07:20:02 MiniClops kernel: ahci 0000:01:00.0: setting latency timer to 64 (Drive related) Feb 1 07:20:02 MiniClops kernel: scsi6 : ahci (Drive related) Feb 1 07:20:02 MiniClops kernel: scsi7 : ahci (Drive related) Feb 1 07:20:02 MiniClops kernel: ata7: SATA max UDMA/133 abar m8192@0xfe8fe000 port 0xfe8fe100 irq 16 (Drive related) Feb 1 07:20:02 MiniClops kernel: ata8: SATA max UDMA/133 abar m8192@0xfe8fe000 port 0xfe8fe180 irq 16 (Drive related)