December 28, 200718 yr So I've been running unRAID basic (Ver 4.0 and then Ver 4.2.1) for about 4 months with no problems. Then I got a gigabit switch (Netgear GS605) for Christmas, and when I plugged the unRAID server into it, I can access it normally via the web interface or windows file explorer for a few minutes, and then it disappears, and won't come back. When I log onto the console, there doesn't seem to be an IP address anymore. If I switch it back to the 100Mb switch, it is still lost until I reboot from the console. Then it comes up fine. I've got windows machines plugged into the switch that are communicating fine at 1Gb, and they don't have any connectivity problems to the unRAID server or the web. Even when the unRAID server is the only line except the uplink in the gigabit switch, it doesn't stay connected. I wanted to try switching back to unRAID 4.0 just to see if it would help, but I'm unsure if I can safely just copy these four files over and not run into problems by downgrading? bzimage bzroot readme.txt syslinux.cfg I don't know if downgrading is barking up the wrong tree, either... should I check something else first? I'm using the built in MB Gb ethernet port - Marvel 8056 Gigabit LAN Controller. The motherboard is the Gigabyte GA-965GM-S2.
December 28, 200718 yr So I've been running unRAID basic (Ver 4.0 and then Ver 4.2.1) for about 4 months with no problems. Then I got a gigabit switch (Netgear GS605) for Christmas, and when I plugged the unRAID server into it, I can access it normally via the web interface or windows file explorer for a few minutes, and then it disappears, and won't come back. When I log onto the console, there doesn't seem to be an IP address anymore. If I switch it back to the 100Mb switch, it is still lost until I reboot from the console. Then it comes up fine. I've got windows machines plugged into the switch that are communicating fine at 1Gb, and they don't have any connectivity problems to the unRAID server or the web. Even when the unRAID server is the only line except the uplink in the gigabit switch, it doesn't stay connected. I wanted to try switching back to unRAID 4.0 just to see if it would help, but I'm unsure if I can safely just copy these four files over and not run into problems by downgrading? bzimage bzroot readme.txt syslinux.cfg I don't know if downgrading is barking up the wrong tree, either... should I check something else first? I'm using the built in MB Gb ethernet port - Marvel 8056 Gigabit LAN Controller. The motherboard is the Gigabyte GA-965GM-S2. You only need to copy bzimage and bzroot to downgrade. I would check to see if the motherboard has any bios updates that may affect the LAN controller. Are you using DHCP? or hard coding IP addresses? Other than that, try pugging the server into a different port on the switch. Try a different cable. Higher speed requires better cabling, with better attention to the twists in the cable and cable quality. (Cat 5 or 6 os OK, Cat 3 is probably not good enough at 1Gig speeds) Joe L.
December 28, 200718 yr Author You only need to copy bzimage and bzroot to downgrade. I would check to see if the motherboard has any bios updates that may affect the LAN controller. Are you using DHCP? or hard coding IP addresses? Other than that, try pugging the server into a different port on the switch. Try a different cable. Higher speed requires better cabling, with better attention to the twists in the cable and cable quality. (Cat 5 or 6 os OK, Cat 3 is probably not good enough at 1Gig speeds) Joe L. Ah, easy to downgrade - unfortunately, it still fails after a few minutes for ver 4.0, so no easy fix there. Odd aside, my data drive (drive 1, only 2 drives in the array right now, parity and drive 1) came up as missing under ver 4.0. I upgraded to 4.2.1, and the array is back and healthy. As far as the bios, I'm running the latest version, and none of the changelogs for the previous versions indicate any LAN related changes (but I'm sure they're not putting everything in the changelogs). I was using DHCP, switching to a hardcoded IP didn't keep it running, so I'm switching back. All the cables I'm using are Cat 5 and Cat5e, even on the 100Mb switch... I'll try swapping out a few just to see, and I'll try all the other ports on the Gb switch as well.
December 28, 200718 yr Author All the cables I'm using are Cat 5 and Cat5e, even on the 100Mb switch... I'll try swapping out a few just to see, and I'll try all the other ports on the Gb switch as well. Well, I started by trying other ports on the switch, (easier than switching out cables), and when I tried the second port, everything is now stable. I've done massive file copies both ways, browsed directories with thousands of files, and madly clicked about on the web interface, and there's not a hiccup. So, I'm both extremely please that it's working (yeah for faster syncs at nights!), and concerned that selecting the right port on the switch could be so critical. Is this a normal problem with Gb ethernet, to have to switch ports? Anyhow, thanks so much for your quick and thorough response, Joe! You made my day!
December 28, 200718 yr I've had ports go flaky on switches and hubs for years.... both high-end Cisco and Bay networks stuff, and el-chepo brands. It happens.
December 28, 200718 yr Author I've had ports go flaky on switches and hubs for years.... both high-end Cisco and Bay networks stuff, and el-chepo brands. It happens. I would expect some failure over time, but right away? Apparently so. Anyhow, now I know.
December 28, 200718 yr in the department of longshots, check the physical port on the switch. The little copper wires, if some are a little "bend up", the connectivity can be bad. In the days before wifi-allover, the lan port on a real roadwarrior-laptop took a some abuse. If the connection gets good when pushing in the cable or shaking it a bit, the physical connection is a good suspect. It can be fixed using a straightend paperclip, where you bend the last 3mm of clip, and use it to get behind the connectors and bend them down. Be carefull, and turn off the switch :-)) /Rene
December 28, 200718 yr If convenient to do so, return the switch. It may not bother you now to avoid one port, but at some point you may need that port and by then it will be too late to return it. Bill
December 31, 200718 yr I've been using that same switch for about 9-10 months, and just recently started having problems like you are describing. I also just recently upgraded from Unraid 4.0 to 4.2, and to top it off I swapped out motherboards. And to top it off, my media pc just started flaking out on me with extremely slow network transfers, which looks like a bad NIC, but maybe it's related to the switch as well (although I did try switching cables and ports with no improvement). But specifically, my Unraid server is randomly losing connection to the network. Over the last 2 days it has happened in the following scenarios: 1) Watching a movie and it lost connection 3 times during a 2 hour movie and had to reboot the server. 2) While copying data to my media pc (to test the transfer issues I'm having with it) 3) While running a telnet session to run reiserfsck on all the drives (which all checked out good, along with a good parity check) So the variety of failures has led me to believe it is a network issue, and I was actually on here to see if there are any problems with my NIC card (Realtek 8168, which appears to be one of the "good" Realteks). I'll try moving everything directly to my 10/100 router ports and see if that fixes the Unraid issue. Hopefully it does, because I was wanting to get an 8 port gig switch anyway and this will help me justify that upgrade. But if this switch is bad, no more Netgear products for me. I also had a wireless router from them that failed after about a year of use.
December 31, 200718 yr Does that switch have a dedicated uplink port? Maybe if you were plugged into it? If not, I agree, return it...
December 31, 200718 yr hmmm netgear here too... also one port seems to work ok only after I remove and reinsert the cable also was planning to get read of it and get their 1Gbit 8 port (I have the 5 port) but they are dead cheap
December 31, 200718 yr Well, I removed the switch from the equation and plugged everything directly into my router and - same problem. At this point I really suspect it is something with the 4.2.1 version, because I copied terabytes of data when setting up the server, and never had any problems. All the problems started after I upgraded from 4.0 to 4.2.1. I tried monitoring the sys log from a telnet session, and no errors were shown, but then it occurred to me that if the error really is something with the network, then telnet probably couldn't update when the crash occurred. So I'm testing again with a monitor hooked up to the Unraid server so I can watch the syslog. Hopefully we can find an easy fix to what's going on, otherwise I need to roll back to the 4.0 version for now. With everyone off school/work right now, I'm getting a lot of pressure to fix the media server, and the WAF is dropping fast.
December 31, 200718 yr Yep, I had to roll back to 4.0 and that fixed everything, including my media pc's read problem. I don't really use any of the new features in 4.2 anyway, so I'll just stick with 4.0 for awhile.
December 31, 200718 yr I rolled back from 4.2.1 to 4.0. 4.0 seems much more stable to me. Does your motherboard also have a Marvell NIC?
January 1, 200818 yr I rolled back from 4.2.1 to 4.0. 4.0 seems much more stable to me. Does your motherboard also have a Marvell NIC? Nope: root@mediaserver2:~# dmesg | grep eth [ 81.766308] eth0: Identified chip type is 'RTL8110SC'. [ 81.766313] eth0: r10001.05, the Linux device driver for Realtek Ethernet Controllers at 0xa800, 00:1a:4d:27:13:8a, IRQ 21
January 1, 200818 yr Neither is mine: Tower login: root Linux 2.6.20. root@Tower:~# dmesg | grep eth [ 86.443797] eth0: Identified chip type is 'RTL8168B/8111B'. [ 86.443802] eth0: r10001.05, the Linux device driver for Realtek Ethernet Con trollers at 0x9e00, 00:50:8d:9d:ba:19, IRQ 17 root@Tower:~#
January 1, 200818 yr both are realtek, which has had reported problems. The driver was either removed or changed on one of the updates if I recall. There are other posts about RealTek issues.
January 2, 200818 yr Hi there! Just for info and if it might help : I also have Realtek NIC (PCI card, not the same model though...) on 4.2.1 and no problem with it. Linux 2.6.22.5. [ 100.889664] eth0: RTL8169s/8110s at 0xf8816c00, 00:A1:B2:C3:D4:E5, IRQ 11 [ 116.850880] r8169: eth0: link up Fixed IP, Gb router is Dlink DGL-4300, CAT6 cables ... ... BTW, when you set up your fixed IP did you put your local DNS in the resolv.conf?
January 5, 200818 yr Author Does that switch have a dedicated uplink port? Maybe if you were plugged into it? If not, I agree, return it... Yeah, it has no dedicated uplink (auto uplink on all ports). All the other ports look fine, and work fine when the Windows machines are plugged into them. They are only a problem for the unRAID server. Post-fix, I've tried every other combination, except I'm leaving the unRAID server on the port it likes, and everything ticks along great. Only when I move the unRAID port do I get problems. Ah well. It was a gift, purchased over the internet for me, so the trouble of returning and whatnot for something that may just as well happen on the next switch is not attractive to me. If I knew why it was happening, I might return it with more confidence...
January 7, 200818 yr Just spitballing here, but does unraid have ethtool on it? May be something in it's output that helps (connect to one port, bump the switch, test, move port, bump the switch, retest). It could be that it's trying to put itself into half duplex or something goofy like that. I've had problems with some switches getting confused when I move ports and I have to reboot the switch (it's internal routing gets messed up).
January 7, 200818 yr I just typed ethtool and this happened... guess it does have ethtool Joe L. root@Tower:/boot# ethtool ethtool version 4 Usage: ethtool DEVNAME Display standard information about device ethtool -s|--change DEVNAME Change generic options [ speed 10|100|1000 ] [ duplex half|full ] [ port tp|aui|bnc|mii|fibre ] [ autoneg on|off ] [ phyad %%d ] [ xcvr internal|external ] [ wol p|u|m|b|a|g|s|d... ] [ sopass %%x:%%x:%%x:%%x:%%x:%%x ] [ msglvl %%d ] ethtool -a|--show-pause DEVNAME Show pause options ethtool -A|--pause DEVNAME Set pause options [ autoneg on|off ] [ rx on|off ] [ tx on|off ] ethtool -c|--show-coalesce DEVNAME Show coalesce options ethtool -C|--coalesce DEVNAME Set coalesce options [adaptive-rx on|off] [adaptive-tx on|off] [rx-usecs N] [rx-frames N] [rx-usecs-irq N] [rx-frames-irq N] [tx-usecs N] [tx-frames N] [tx-usecs-irq N] [tx-frames-irq N] [stats-block-usecs N] [pkt-rate-low N] [rx-usecs-low N] [rx-frames-low N] [tx-usecs-low N] [tx-frames-low N] [pkt-rate-high N] [rx-usecs-high N] [rx-frames-high N] [tx-usecs-high N] [tx-frames-high N] [sample-interval N] ethtool -g|--show-ring DEVNAME Query RX/TX ring parameters ethtool -G|--set-ring DEVNAME Set RX/TX ring parameters [ rx N ] [ rx-mini N ] [ rx-jumbo N ] [ tx N ] ethtool -k|--show-offload DEVNAME Get protocol offload information ethtool -K|--offload DEVNAME Set protocol offload [ rx on|off ] [ tx on|off ] [ sg on|off ] [ tso on|off ] [ ufo on|off ] ethtool -i|--driver DEVNAME Show driver information ethtool -d|--register-dump DEVNAME Do a register dump ethtool -e|--eeprom-dump DEVNAME Do a EEPROM dump [ raw on|off ] [ offset N ] [ length N ] ethtool -E|--change-eeprom DEVNAME Change bytes in device EEPROM [ magic N ] [ offset N ] [ value N ] ethtool -r|--negotiate DEVNAME Restart N-WAY negotation ethtool -p|--identify DEVNAME Show visible port identification (e.g. b linking) [ TIME-IN-SECONDS ] ethtool -t|--test DEVNAME Execute adapter self test [ online | offline ] ethtool -S|--statistics DEVNAME Show adapter statistics ethtool -h|--help DEVNAME Show this help root@Tower:/boot#
January 7, 200818 yr Do any of you have multiple nics in your box? In my testing, I have had unRAID grab different nics on different boots. I have a Tyan board with 2 built in 10/100 ports, and a PCI realtek GigE card. One time unRAID grabbed the GigE card, but now it seems to want to grab one of the 10/100 ports all of the time. I have the built-in motherboard ethernet disabled in BIOS, but Linux doesn't seem to care about that. I have tryed booting up with the cable in different ports, and can't reproduce the GigE card getting chosen.
Archived
This topic is now archived and is closed to further replies.