This topic has been marked solved and closed to new posts due to inactivity. We hope you'll join the conversation by posting to an open topic or starting a new one.
Well, I followed the instructions from the other 120+ page thread and upgraded my Pro-nas Pioneer from 4.27 to 6.3.5, which BTW takes <somewhat> longer than 10 minutes on a 12 TB system - about 4hrs 36min longer - but the process was pretty painless and worked as detailed in the postings with no surprises.
After the second reboot, the status page proudly announced that there was an upgrade to version 6.4.0 available, and since I'd not yet reloaded the drives, I thought "What could possibly go wrong?' and accepted the upgrade.
While I'm pretty confident that there's a lot which could have gone wrong - Netgear apparently employs union gremlins who clock out promptly at 18:00 - because the upgrade installed swiftly and without any obvious errors and seems quite stable in my pre-production testing.
For those who care about these things, the box has been mildly upgraded to 4Gb of Crucial DDR2, and a E7600 Core Duo, the combination of these upgrades and the newer Linux distro has made the system a very effective home server - it has no problems dealing with an almost 2Tb music library, distributing via Logitech Squeezebox multiple audio streams, and appears to be an effective backup solution for several Windows 10 machines.
I do have squawks out on two system anomalies/bugs:
The first bug was relatively trivial to patch, but may indicate deeper problems with the way that Netgear has choosen to 'harden' Debian's dpkg & apt-get package management system. The executable 'leafp2p' had a library conflict with the version of Logitech Server offered from Netgear, which created difficulties in installing some of the Server's optional plug-ins.
Since I don't intend to use any of Netgear's system remote offerings, I disabled leafp2p, removed its scripts from init.d, and with it gone from memory the install conflicts are no longer a factor - although I remain concerned that Netgear's install process apparently breaks the distro.
The second anomaly appears to be a pretty simple logic glitch in the routines that correlate the CPU/System temperatures and the fan speed controllers. With the case closed, the CPU is stable at 39C (105F), but the system temp gradually creeps up to 70C (160F), while the fans DROP in speed from 1400RPM to 1180 (CPU), and from 880 to 635RPM (System).
If, OTOH, I remove the case side, the CPU drops to 32C and its fan increases to 1980RPM, while the System drops to 54C,, but its fan increases to 1094RPM - and yes, I've watched the fans and their sensors appear to be correctly reporting their speeds.
This looks like the sort of classic 'Bass-ackwards' result that happens when a variable gets inverted or a flag mis-set, and should be fixable.
I gather that some folks here have done significant hacking with the fan code (and I'm also assuming that these routines are Netgear closed source) but could someone point me to the correct place to begin looking for a patch-point?
Do not upgrade to 6.4.0 unless you want major headaches. My pro 6 and ultra 6 tell a story of random drives dropping out of the volume and lots of lockups and just other odd/bad behavior. I don't think 6.4.0 was ready for primetime so I wish it wouldn't have prompted me to do so. I should just switch back to 4X and be done with it.
Reply to my message of 10/09/2015 with a 14-day update:
After two wweks of continuous running with no crashes, failures, or unusual drive/volume issues, I'm about ready to put this server into 'production' status - which will move it from my desktop down to a somewhat inacessable root cellar off of our basement.
The only thing stopping this move are my concerns about the misconfigured fan speeds causing possible thermal instability - since if either the system controller or the CPU begins to warm up, the inverted controller logic slows the fans down - which strikes me as a receipe for thermal runaway.
I've been experimenting with opening the side case next to the system IC heatsink and the CPU fan, and have found that cracking the case open 1/4-1/3" and then holding it in place with a strip of 3M 'Micropore' breathable tape, yields consistant CPU temps of 102-101F (39C), and system temps of 150-152F (52C) with accpetable fans speeds of 1635 (CPU) and 835 (System) all of which seem quite stable under loads such as large CIFS copies and Squeezebox Database scans which typically show 80-95% CPU utilization under HTop.
These are encouraging results, but I'd prefer a solution that's just slightly more sophisticated than 'Dreadful Hack,' before consigning the server to the cellar.
I've also done repeated runs of the RaidIR volume utilities (Defrag, Scrub, and Balance) on the server's 12TB Raid 5 array, with no hiccups or unexpected results, and nmow have a fair degree of confidence that the drives and volume are stable under load and actual use.
I'd read ifixidevice's comments with some trepidation, but now feel that his drive issues may be indicative of other problems in his hardware environment that the more recent Debian kernel is exposing - absent any specific information from him, or creditable reports from other users, I'll assume I'm either very fortunate, or correctly configured, or (for a change) both...
Not sure how I don't have my box configured right but oh well. I've got a Q6600 in the device and have 2GB's of memory. I get frequent lockups over periods of time (of course I added a hard drive and have added a few and during the resync process the thing likes to lock up if you put any load on it.
It locked up this weekend while it was doing the following... resyncing, being used as an SMB server for a mac share, copying files from it as an SMB share on a mac to ftp onto another nas box. Was also doing a time machine backup. All things I'd imagine it should be able to do and did so very well on the 6.2.4 software.
My RN104 seems to be mostly stable which is a miracle given that so many people are complaining about that unit having issues with 6.4.0
Basically look on the forums and see all the issues that 6.4.0 is having on so many different current and legacy devices. It feels very unpolished and glitchy. Basically using the same setup (minus a few drives) on 6.2.4 I had no issues whatsoever and the box never locked up at all. With 6.4.0 I know at some point it will lock up (hard lockup) and I'll have to restart it.
Sorry, I didn't mean to imply that you hadn't correctly setup your NAS - but rather that I've a hunch that you're having either process buffer (memory) or cache coherency (CPU) issues - which the Debian 4.7.2-5 Wheezy kernel appears to be exacerbating - this based on your statements that the NAS ran well until the upgrade to firmware 6.4 and your ARM-based server runs 6.4 satisfactorilly
Correct me, but I believe that you're running an Intel Q8800 quad-core in a ReadyNAS Pro with 2 Gb of memory and a fair amount of storage?
My hunch - and this is <only> a poorly-documented hunch - is that you've run afoul of an architecural flaw in the last generation of core-duo processors, which weren't designed as quad-core devices, and whose cache could get swamped and/or overwritten if there were to many concurrent interrupts for the CPU to handle. This issue was addressed in the Nehalem/I3 dies, but wasn't seen as particularly critical since neither Windows 7, MS Server 2000, nor the Linux 2.6.x kernel of 2008 - were sufficently optimised to drive these quad-cores into saturation.
Now, however, "The Fooh is on the other Shoot.." {Dr Erwin Corey) and it looks like Netgear's fork of the md/LVM driver is pretty heavily optimized for mult-core performance, and might well be driving the Q8800 too hard - especially if the 2Gb of memory isn't enough to provide adequate buffering for multiple drivers.
The relatively cheap and easy way to test this would be to drop back to a true core duo like an E7600 - they're readily available and cheap (I paid $25 for mine from eBay seller Sunnking), pop right in, have a lower thermal load than the stock CPU's, give respectable performance at about 6120 Bogomips, and I've had no apparent lockups or random glitches with this install.
Not sure how I don't have my box configured right but oh well. I've got a Q6600 in the device and have 2GB's of memory. I get frequent lockups over periods of time (of course I added a hard drive and have added a few and during the resync process the thing likes to lock up if you put any load on it.
It locked up this weekend while it was doing the following... resyncing, being used as an SMB server for a mac share, copying files from it as an SMB share on a mac to ftp onto another nas box. Was also doing a time machine backup. All things I'd imagine it should be able to do and did so very well on the 6.2.4 software.
My RN104 seems to be mostly stable which is a miracle given that so many people are complaining about that unit having issues with 6.4.0
Basically look on the forums and see all the issues that 6.4.0 is having on so many different current and legacy devices. It feels very unpolished and glitchy. Basically using the same setup (minus a few drives) on 6.2.4 I had no issues whatsoever and the box never locked up at all. With 6.4.0 I know at some point it will lock up (hard lockup) and I'll have to restart it.
