limetech Posted November 27, 2013 Share Posted November 27, 2013 Download Please see readme.txt in the release zip file for installation/upgrade instructions. I thought I'd expand a bit on the changes from 5.0.2 to 5.0.3: - emhttp: avahi smb service name changed from "%h-SMB" to "%h" (%h is your server hostname) - emhttp: avahi afp service name changed from "%h" to "%h-AFP"; and generate proper time-machine records These are changes to how the SMB and AFP services are broadcast via AVAHI and how this interacts with the OSX 'finder'. It seems that 'finder' will list servers it finds both via netbios and via avahi. In unRaid versions prior to 5.0, we did not enable avahi unless AFP was also enabled. In this case, the service name for AFP was set to the server hostname (e.g., "tower") and the service name for SMB was set to the server hostname with a "-SMB" suffix. When finder sees the same netbios name as avahi name, it displays the avahi name and clicking on it lets you browse the server using AFP - and clicking on the finder entry for server name-SMB lets you browse the server using SMB protocol. In 5.0 we made a change to always enable avahi, even if AFP was not enabled. The reason for this was to make it easier for an OSX user to get to the webGui by typing "tower.local" in the browser address bar, rather than IP address. The side effect was that both "tower" and "tower-SMB" showed up in the 'finder' side bar. The entry for 'tower' is the netbios name and 'tower-SMB' is our avahi service name. Since we can't make 'finder' display something besides the netbios name for netbios name, I changed the avahi service definitions so that in 5.0.3, it's the AFP service name with a suffix, e.g., 'tower-AFP'. Anyway some users will like this, other maybe not. This can be configured using these variables in the /boot/config/share.cfg file: shareAvahiEnabled default "yes" shareAvahiSMBName default "%h" shareAvahiSMBModel default "Xserve" shareAvahiAFPName default "%h-AFP" shareAvahiAFPModel default "Xserve" So if you want to change it back the way it was you would add these lines in /boot/config/share.cfg: shareAvahiSMBName="%h-SMB" shareAvahiAFPName="%h" The 'Model' strings define which icon 'finder' will display. Look here: http://simonwheatley.co.uk/2008/04/avahi-finder-icons/ Note: I did not hook up any webGui controls for this, you will have to reboot your server for changes to take effect, sorry. The controls will be added to the 'development' webGui. The other change to avahi was to include the proper txt-records for time-machine volumes. - linux: patch mvsas driver to recognize newer AOC-SAS2LP-MV8 cards with PCI ID 1b4b:9485 This one was a bugger. We received a new batch of these cards for a production run of AVS-10/4 servers, and first unit, no drives attached to the card were recognized - oh well sometimes we get bad hardware. We then determine all the cards don't work - oh crap. Well it turns out Supermicro went and changed the PCI ID for the raid-less model cards (why? who knows). Simple patch to fix it. - linux: update Realtek-supplied r8168 driver to version r8168-8.037.00 A user pointed out a new update, so hey why not? - linux: added numerous additional options to netfilter/ip tables subsystems per user request Per user request. - netatalk: upgrade to version 2.2.5 This is probably the last of the netatalk 2.x series. unRaid 6 includes netatalk 3.x. - slack: added package nss-mdns-0.10 and modified /etc/nsswitch.conf to resolve *.local hostnames This gets rid of a avahi warning syslog message, and lets you lookup hostnames in the .local TLD via unRaid console. - unraid: permit "md_write_method" to be set via mdcmd Ok, this is actually an interesting feature possibility. After you Start your array, from the unRaid console or telnet session, you can type this command: mdcmd set md_write_method 1 This will make the unraid driver use "reconstruct-writes" for all write operations to the array. You can change it back to normal "read-modify-write" by typing: mdcmd set md_write_method 0 or Stop/Start the array. This setting has no webGui control and is not persistent across array Stop/Start, and should be considered "experimental" - meaning, your data will be safe, but whether or not you gain any benefit from the setting will depend on your array width and specific hardware. In addition, this setting will cause all disks to be spun up for all writes. For small arrays, PCI-E controllers, there could be a significant (2x) speed up in writes (limited by network). As array width increases, speed benefit decreases, and at some point will be slower than normal read-modify-write method. What I'm interested in are user reports of whether you find this setting beneficial. Please include your h/w details in your report. I'm thinking I might hook up a webGui control like this: if this "reconstruct-write" mode is enabled, then if the s/w detects all drives are spun up, it will do reconstruct-write, else if any non-target disk is spun down it will do normal read-modify-write. Quote Link to comment
jumperalex Posted November 27, 2013 Share Posted November 27, 2013 nice work Tom. Quote Link to comment
garycase Posted November 27, 2013 Share Posted November 27, 2013 Nice job => I haven't tested the new write option, but your idea of making this a "do it this way if all the disks are spun up; else do read/modify/write" seems like a nice enhancement. Quote Link to comment
gfjardim Posted November 27, 2013 Share Posted November 27, 2013 Thank you Tom for Avahi and Netatalk enhancements! Quote Link to comment
nars Posted November 27, 2013 Share Posted November 27, 2013 Thank you, will test md_write_method soon. Quote Link to comment
limetech Posted November 27, 2013 Author Share Posted November 27, 2013 Thank you Tom for Avahi and Netatalk enhancements! Thank you for your previous input, much appreciated! Quote Link to comment
Finka Posted November 27, 2013 Share Posted November 27, 2013 Could have as well upgraded Samba to the latest in the 3.6.x series. (3.6.20 as of now). For me personally Samba is the most important part of unRAID. Quote Link to comment
pantner Posted November 27, 2013 Share Posted November 27, 2013 Thanks Tom. FYI, still having downloading issues. Just downloaded on a W8 machine in IE11, the download was 'interrupted' once, then when it looked like it finished the zip file was corrupted. currently trying Chrome on the same computer. *EDIT - Chrome did even worse. IE downloaded ~11MB. Chrome downloaded ~3MB. Quote Link to comment
limetech Posted November 27, 2013 Author Share Posted November 27, 2013 Could have as well upgraded Samba to the latest in the 3.6.x series. (3.6.20 as of now). For me personally Samba is the most important part of unRAID. Agreed. Focus with 5.0.3 was to address some issues with OSX integration, as well as fix major issue with supporting newer revisions of the AOC-SAS2LP-MV8, so I had to get this out. Quote Link to comment
binhex Posted November 27, 2013 Share Posted November 27, 2013 Thanks Tom. FYI, still having downloading issues. Just downloaded on a W8 machine in IE11, the download was 'interrupted' once, then when it looked like it finished the zip file was corrupted. currently trying Chrome on the same computer. *EDIT - Chrome did even worse. IE downloaded ~11MB. Chrome downloaded ~3MB. I too had issues with chrome and ie, ended up using wget via telnet session on unraid and it worked first time so might be worth a shot. Quote Link to comment
PeterB Posted November 27, 2013 Share Posted November 27, 2013 I've just downloaded this, using wget because recent experience had shown that fetches were running slow and getting interrupted. However, the fetch worked flawlessly - no interruption and a reported speed of 182kB/s (I'm on a 2Mb connection here in Philippines). Perhaps I was just lucky, or maybe the problem has been resolved? Quote Link to comment
Harpz Posted November 27, 2013 Share Posted November 27, 2013 Great work Tom, will test shortly, any new on when the new GUI will be getting and update? Quote Link to comment
spylex Posted November 27, 2013 Share Posted November 27, 2013 Great work, thank you. As other users have also mentioned, bringing all the important linux packages up to date will be amazing! Quote Link to comment
S80_UK Posted November 27, 2013 Share Posted November 27, 2013 Updated to 5.0.3 without issues. Thanks Tom. Happy to report that the minor Avahi changes did not cause any problems with the Firefly server program that I run. Still streaming music to my SoundBridges no problem. Quote Link to comment
EdgarWallace Posted November 27, 2013 Share Posted November 27, 2013 All is working well - thanks a lot for all the great enhancements. A minor item, that I would like to mention: Time Machine doesn't find the former backup anymore, respectively it doesn't have access. I guess, that it has something to do with these change? - emhttp: avahi smb service name changed from "%h-SMB" to "%h" (%h is your server hostname) - emhttp: avahi afp service name changed from "%h" to "%h-AFP"; and generate proper time-machine records Anyone having the same problem or is there something that I overlooked? Quote Link to comment
Interstellar Posted November 27, 2013 Share Posted November 27, 2013 You will have to remove it from the disk list and re-add it. Reason being that the path to the backup has changed, but it will find it again if you point it in the right direction. Quote Link to comment
gfjardim Posted November 27, 2013 Share Posted November 27, 2013 Great work! Hopefully you will get a chance to look into SCSI virtual disks (ESXi) for next version of Unraid There's nothing Tom could do; or ESXI exports disk serial number, so udev rules can create the appropriate symlinks, or not. For Tom to address this from the unRAID side, it would involve using other attributes, maybe partition UUID, but this would be a hack. Just set your VM to enable disk UUID. Quote Link to comment
binhex Posted November 27, 2013 Share Posted November 27, 2013 I've just downloaded this, using wget because recent experience had shown that fetches were running slow and getting interrupted. However, the fetch worked flawlessly - no interruption and a reported speed of 182kB/s (I'm on a 2Mb connection here in Philippines). Perhaps I was just lucky, or maybe the problem has been resolved? PeterB, this has been my experience too, every time i use wget it just works!, if i use chrome or ie it generally fails, its odd but hey for now its a workaround :-). btw just in case wget does start failing you can use the -c option to resume :-) Quote Link to comment
peter_sm Posted November 27, 2013 Share Posted November 27, 2013 I have on my mac tower tower-AFP But both these 2 are connected to my AFP share, so no SMB share shows up in finder ....... thsi is some rows in the syslog Nov 27 16:16:02 tower avahi-daemon[1491]: Server startup complete. Host name is tower.local. Local service cookie is 775833280. Nov 27 16:16:03 tower avahi-daemon[1491]: Service "tower" (/services/smb.service) successfully established. Nov 27 16:16:03 tower avahi-daemon[1491]: Service "tower-AFP" (/services/afp.service) successfully established. Nov 27 16:18:03 tower afpd[1701]: dsi_stream_read: len:0, unexpected EOF Nov 27 16:19:42 tower afpd[1712]: dsi_stream_read: len:0, unexpected EOF //Peter Quote Link to comment
EdgarWallace Posted November 27, 2013 Share Posted November 27, 2013 Peter_SM, I do have the same effect. My initial thought was, that is has something to do with the launchd/mount script? http://lime-technology.com/forum/index.php?topic=10274.150 Quote Link to comment
gfjardim Posted November 27, 2013 Share Posted November 27, 2013 I have on my mac tower tower-AFP But both these 2 are connected to my AFP share, so no SMB share shows up in finder ....... thsi is some rows in the syslog Nov 27 16:16:02 tower avahi-daemon[1491]: Server startup complete. Host name is tower.local. Local service cookie is 775833280. Nov 27 16:16:03 tower avahi-daemon[1491]: Service "tower" (/services/smb.service) successfully established. Nov 27 16:16:03 tower avahi-daemon[1491]: Service "tower-AFP" (/services/afp.service) successfully established. Nov 27 16:18:03 tower afpd[1701]: dsi_stream_read: len:0, unexpected EOF Nov 27 16:19:42 tower afpd[1712]: dsi_stream_read: len:0, unexpected EOF //Peter Will investigate this soon. Have you rebooted your Mac? Maybe Finder cached some advertised parameters. Quote Link to comment
peter_sm Posted November 27, 2013 Share Posted November 27, 2013 I did disable / enable the AFP and SAMBA, now all good Quote Link to comment
limetech Posted November 27, 2013 Author Share Posted November 27, 2013 I did disable / enable the AFP and SAMBA, now all good Right I saw some funny things in finder during testing that usually were solved by just relaunching finder. Quote Link to comment
peter_sm Posted November 27, 2013 Share Posted November 27, 2013 I did the restart of samba / afp on the server side My MAC connected fine again Quote Link to comment
EdgarWallace Posted November 27, 2013 Share Posted November 27, 2013 I did the same - worked for me as well. Thanks to Peter Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.