Jump to content

[SOLVED] Writing to share has slowed to a crawl


Recommended Posts

I replaced my 1tb parity drive with a new 2tb drive because the 1tb drive failed (lots of write errors). After the initial parity sync was complete, I tried copying some new movies to the server and it was stupidly slow - like 200kb/sec.  I looked in the syslog and saw some acpi errors. I updated the bios but they werent fixed so I after searching here, I addeed ACPI=OFF in the cfg file.  After a reboot, no more syslog errors but the copy to the server is still the same super slow speed.

 

Any ideas??

 

I am running 4.5.6.  I have three 1tb drives and the 2tb as the parity.

DG41BI / Intel E5500 2.8Ghz / 4 gb ram

 

Thanks!

 

ps. I started another parity check.  It's running at about 34,500 kb/sec.

 

syslog.txt

Link to comment

Parity drive: Western Digital Caviar Black WD2001FASS 2TB 7200 RPM 64MB Cache SATA 3.0Gb/s - no jumpers need to be set (sdc)

 

Data Drives: Seagate Barracuda 7200 1 TB 7200RPM SATA 3Gb/s 32MB Cache  - no jumpers need to be set (sda, b, d)

 

Drive sdd seems slow on the timing buffered disk reads:

 

hdparm Test:

 

/dev/sda:

Timing cached reads:   2244 MB in  2.00 seconds = 1121.56 MB/sec

Timing buffered disk reads:  368 MB in  3.01 seconds = 122.18 MB/sec

 

/dev/sdb:

Timing cached reads:   2208 MB in  2.00 seconds = 1103.68 MB/sec

Timing buffered disk reads:  366 MB in  3.01 seconds = 121.68 MB/sec

 

/dev/sdc:

Timing cached reads:   2254 MB in  2.00 seconds = 1126.60 MB/sec

Timing buffered disk reads:  406 MB in  3.01 seconds = 135.07 MB/sec

 

/dev/sdd:

Timing cached reads:   2194 MB in  2.00 seconds = 1097.16 MB/sec

Timing buffered disk reads:   42 MB in  3.03 seconds =  13.88 MB/sec

 

smarta.txt

smartb.txt

smartc.txt

smartd.txt

Link to comment

I think I am on to fixing the problem.  I ran the memory test and bang - the second and third tests failed.  New memory is due tomorrow. Hopefully that will fix the issue.  This memory is only 30 days old. Tiger direct won't RMA it - have to deal with the manufacturer directly.  Thats just wrong.

 

Nice build, within 30 days 1 hard drive goes and 1 memory ddr3 turns out to be bad.  Whats next, the mobo?

 

I will update this thread as information warrants as it may help someone else who experiences the same issues.

Link to comment

I think I am on to fixing the problem.  I ran the memory test and bang - the second and third tests failed.  New memory is due tomorrow. Hopefully that will fix the issue.  This memory is only 30 days old. Tiger direct won't RMA it - have to deal with the manufacturer directly.  Thats just wrong.

 

Nice build, within 30 days 1 hard drive goes and 1 memory ddr3 turns out to be bad.  Whats next, the mobo?

 

I will update this thread as information warrants as it may help someone else who experiences the same issues.

The memory might not be bad... It might be that the BIOS did not proprtly set the timing, voltage and clock speed for your specific RAM strips.

Most modern BIOS get ti right, but an awful lot do not, especially with premium RAM that needs special timing and/or voltage/or clock speed.

 

Glad you figured it out though, since memory errors can lead to hair loss.  (You'll pull your hair out trying to find the cause of elusive random errors)

 

Joe L.

Link to comment

An update - I still have a problem writing to the share.

 

I have done 2 things since my last post. I changed the power supply - to a 900w from a 450w.  Changed the defective memory with new memory so now I am at 4gb.  Still copying files like a ripped dvd to the server just crawls.  

 

I also thought maybe the source PC is the problem.  So I went to another PC and tried copying something and it also went very slow. I am getting about 1.4mb/sec when copying to the share and thats using robocopy in windows 7 which worked better than windows explorer.

 

One other note, while copying to the server/share, it sometimes looses the server (robocopy says server not available and waits 30 seconds and then continues) (windows explorer reports that the folder isn't available and then the copy ends with the FAT being updated saying the files are present and their size is correct, but they are not complete).

 

Help?

syslog.txt

Link to comment

No its the hardware.  I replaced the sdd drive which reported that really low speed on the hdparm report and it also reported the following on the latest smartctl report:

 

188 Command_Timeout        0x0032  100  099  000    Old_age  Always  21475164165

 

I took that drive out and put a new one in. It is still rebuilding, but I can already copy to the share at 10mb/sec even while the server is very busy rebuilding.

 

Thats two of the 1tb drives failed within a month (although I think this one was bad from the get-go).  For those of you who have been doing this awhile,  what kind of drive failure rates do you see? Is this just bad luck or should I expect to be replacing a drive per month?

 

Thanks

Link to comment
  • 1 month later...

Two issues. One was my cabling was CAT-5 not CAT-6 but also drives went bad. The cabling issue accounted for running at 100mb instead of 1gb speeds.  The bad drive problems slowed performance down to next to nothing on writing.

 

This original issue was based on a data drive that was reporting errors in the smart report. After replacing the drive, it went back to normal speed. Then it got bad again because my 2tb parity drive started going. Actually over the past couple months I have replaced my parity drive 2 times - I am on my third.  They started reporting sector errors and then I would run the tool from WD and it would fix that, then it would happen again.  My parity drive is the WD Black 2tb 7200rpm 64mb cache drive. I got them at best buy and they have a 30 day warranty so its been easy to take them back to the store and exchange for another. Hopefully this third one will be stable.  So I have had 2 bad data drives (2 of the 4 failed & were replaced) and 2 bad parity drives.

 

Link to comment

To create the FreeDOS USB boot device I downloaded this bootstick file then followed these directions which are very similar to unRAID USB Flash prep.

 

Make sure you're working as administator and assume X: is the usb sticks drive letter

  • Format your USB flash drive using FAT32
  • From a Administrator Dos Prompt invoke syslinux\syslinux.exe -fma X:
  • copy the files underneath "usb-root\" to the root of "X:"

Link to comment

1. Took the ISO from WD and burned to a CD on my windows pc

2. This is the hard part - until I found the solution: downloaded james' format tool (http://www.jamesonline.ca/support/dos-on-usb-support). This program created the boot sector on the flash drive.

3. Copied the dos programs that came with it to the flash drive

4. Copied the WD tool files from the CD to the flash drive

5. Put the flash drive into my raid server instead of the raid flash

6. It booted up and was at a dos prompt

Link to comment

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...