Search results

Good to hear. (Thanks for the follow-up.) That indicates cables had been right/good all along; also that there had been no power supply anomalies in your tower/DAS, which could have explained the symptom, I can not envision that merely bad/wrong firmware in your original cards could explain it...

IF it/they really are reverse cables, is that the expected behavior? (max of one connection) Since a SAS port, and the connected cable, carries 4 physically separate Rx/Tx SAS/SATA connections [see SFF-8088 pinout], that seems odd. In addition, while it's not impossible, I've never heard/seen a...

Superficially, they will look the same; the difference is the pinouts/wire-connections. The description on that product page makes it clear that it IS a reverse breakout cable. If a 50cm long cable will suffice, This is cheaper. Disclaimer: I am not 100% certain that [SAS] reverse-breakout works...

reverse

Thanks for the datapoint.

8x 2.5" SAS/SATA subsystem in 2 x 5.25" bays is sometimes available in the used/surplus scene for ~$100-150 e.g., Intel P/N #FUP8X25S3HSDK ... would take two simple/cheap SFF8643-to-SFF8643 cables from H730p Good luck ...

I looked at the specs for your mobo and ... By using the 2 M.2 omboard connectors, yyour current two 970s are sharing PCIe bandwidth, since they are both using the z390 chipset, which relies on a 4-lane connection to the CPU. [The chipset effectively acts as a PCIe switch to share/mediate all...

I bet you need to enable Write_Caching on the drive (ie, nothing wrong with your enclosure/bridge_chip). (Your read speeds are just fine, right??) If I'm right, you need to ask OMV people WHY!! they are so "insistent" on disabling it (either explicitly, or by deceiving the user to do it [ie...

Make careful note, though, that the ASM2812 is only (PCI g3) x4 (max) upstream. It matters not that the card is "x8". "The sh*t they pull, huh??" [Johnny Utah/"Point Break"]

HPA - Host Protected Area

You got TRIMmed. (The "law" of unintended consequences)

If you've decided that a "server-class" HBA is really your best choice, I think you would be better off with a LSI 9207 (instead of a 9211). The 9207 uses PCIe 3.0 (vs 2.0 on the 9211). That means you can get the same throughput with the 9207 in your x4 slot (x16 physical) as you would get...

Whoa!! The RC-508 has (4) USB 3.0 ports. And, all this time I thought you were asking about [from the OP Subject:] "adding more sata ports". Please clarify. OK I read your subsequent discussion w/kirbyrj, and see that you had "wanted" USB. But, had you already used all 6, or 9, USB3 ports on...

It's all about getting the SATA bandwidth that you need, while making efficient use of the limited PCIe lanes you have available--and, still allow for future/further expansion. Consider, other than your 3 PCIe x1 slots, you only have a single PCIe x4 (lane) [in a x16 slot] available. I wouldn't...

Here's an M.2 adapter. It really does work. [And, it does not utilize a port-multiplier.] 2-Lane M.2 PCI-Express 3.0--JMB585 chipset ===== "Your debutante knows what you need, But I know what you want." --Bob Dylan

I agree with you; there is something interesting at play here. [You've heard the expression: "The devil is in the details.", right?] Well, it is possibly pertinent to know just how you've "achieved" your USB2 connection. I am prompted to ask because your "Seq 1M, Q1T1" numbers for USB2 [44/44]...

All of those 4KB random read numbers are bogus--the X300's insanely so. [Toshiba 3.51 MB/s -- feh !!!! ; HGST 0.97 MB/s -- nah!] Ergo: userbenchmark.com is for lusers. Just as an example: Avg. seek time 8 millisec Avg rotational latency (@7200 rpm) 4.16 millisec [therefore, avg access time =...

In addition to Mount Points (because Drive_Letters are so ... douchey), (re-)familiarize yourself with Symbolic Links. Now that you are completely controlling the physical locations, and hierarchical layout, of your files, SymLinks allow you to create alternate "virtual" layouts/hierarchies...

I remember that Samsung bug/fiasco very well. A drive firmware bug resulting in data loss is a pretty big deal. When Samsung fixed the bug, they made a big show of it--special web page, pictures of affected drive labels, and, of course, the usual file downloads and instructions. But, here's the...

I'm not a gamer (except, maybe, PDP-1 Spacewar ~50 yrs ago), but I know a bit about HDD performance. The new HGST 4TB NAS drive [V2] (HDN726040ALE614) is a surprisingly good performer. Superficially, it differs from the original [V1] 4TB NAS (HDN724040ALE640) by having 128MB cache (vs 64MB for...

[ Nothing has changed since post #20, where I said: ] "No difference." (based on your projected usage description.) -- uhclem

Yes, that is true. But the (legal) regulations are what the company MUST comply with. Company policies, both written and unwritten, are often more permissive, especially if you make your case directly to a more senior person. -- uhclem "Remember, the large print giveth, and the small print...

In (my) post #19 (above), I was replying to (a different member's) post #14. (That member self-identifies as a Seagate employee.) Clear now, I hope ... -- uhclem

No difference. (Maybe I should have written "very very specific requirement" :).) Regarding your followup questions: 1. Your 35MBps speed suggests that you were only getting USB 2.0 performance. 2. When connected internally (SATA2 or SATA3 [no difference (measurable by you)]), you should see no...

Why do you say 5400, when Seagate's own Drive Specifications for this drive (ST4000DM004) state that it is 5000 rpm? (I assume you're in Support ... but) If you have access to Engineering people, please ask them why the Drive Identify command reports 5425rpm, when the spec states 5000rpm (or...

In the PDF document that drescherjm linked to, the Rotational Speed of the ST4000DM004 is "stated" (albeit, indirectly) to be 5000 rpm. (In a similar doc, the ST4000DM005 [ http://www.seagate.com/www-content/product-content/barracuda-fam/barracuda-new/en-us/docs/100804656a.pdf ] is "stated" to...

I believe the latter is the case, just from experiments I've done over the last couple of hours. That might explain the continually-increasing count you are seeing from your testing. Note that it is not unusual to have a few (< 10) such "slow sectors" on a drive. Personally, I don't like to...

Hi Mike, Firstly, thanks for making me aware of the fact that many hard drives maintain a counter of "ReadRecovery Attempts". I wrote a low-level program to detect read-retries and log the location and duration; I've used this to check out all new drives I've bought over the last 5 years. In...

Yes, that is (precisely) the correct method. [drescherjm: Since Dan did not specify a bs=N (blocksize), dd uses its default of 512 (bytes), which is appropriate for this command. (equal to LBA size)] [Hypothesis:] The drive's firmware is flawed--it is not anticipating the case where the...

Run badblocks on Cygwin under Windows.

It is not unusual to see a ~5% variation in sequential performance between drives of the same make/model(/fw). (Firmware version has negligible effect on seq performance.) The seq performance for a particular sample of a drive is totally determined during the factory low-level formatting of...

You should have 2.5) SMART tests or you might lose useful info about drive problems, if you wait for 4). --UhClem

You are wrong! No. [I am a little "conflicted" here--I don't mean to defend SpinRite.--it is obsolete/overhyped] But, it does do more than badblocks. The shortcoming of badblocks is that (true to its name) it only reports bad blocks (sectors). [And, it doesn't even report all of them.]...

For multiple (concurrent) ripping/processing sessions, (even) a single SSD may prove faster. You want to eliminate disk access time, which is where SSDs shine.

Good explanation ... but you might never get any CRC error indication (from ZFS) IF that bit-glitch occurred prior to ZFS starting its protection of that data. (Ie, over the network, or upon its initial write to the ZFS host OS buffer) I believe ZFS has some copy tools to help eliminate this...

I would return it. The fact that there were 4 "very slow sectors" from the factory (presumably) is somewhat concerning. But, you did the right thing by completely re-writing the entire disk, and testing again. Now that there are, not zero "delays", but even more than originally, suggests...

Yes, that was the (rough) gist of my point ... that an Atom might be a significant constraint when it really mattered (a rebuild). I don't use ZFS, but, in addition to the RAID calc, aren't the data checksums also in play during verifications, and a rebuild? The overall point is that...

(many of) You appear to be content with a max of ~100 MB/s (ie, Gbit) of data "bandwidth" to/from your array, as is suggested by an Atom processor being sufficient. Question: Doesn't that imply a drive rebuild time of 30-40+ hours when you have a drive failure (based on a 6-7 [x2TB] drive...

No. It is from word 217 in the data block returned by the ATA command "Identify Device". In the case where a drive chooses not to include that in the data block, or if you want to confirm/verify the number, it can be determined programatically using low-level read timings.

Resist the temptation to go for an enclosure that supports both USB3 and eSata. There is a possibility that your USB3 (with [and w/o] UASP) performance might be limited by the addition of the data multiplexing chip needed to support the dual interface.