NETGEAR is aware of a growing number of phone and online scams. To learn how to stay safe click here.
Forum Discussion
Initial sync after reset very slow
I have a 7 year old Ultra 2 Plus, with 2 x 2TB WD Green drives - this issue is with this box.
I have just got a new Readynass 214 with 4 x 2TB WD Red drives as RAID10 with flexraid. This is working fine, although the initial sync took well over 24 hours.
The "older" Ultra 2 is to be used as a backup system, so I have updated it to OS6.9.4 with the procedure provided by the netgear community. This ends with a full reset and resync of the discs.
I have left the Ultra 2 set to use X-RAID, as with 2 discs its Raid1 anyway.
The resync stated 102 hours to complete, its currently still running with 35% complete and 68 hours to go.
Having searched the community forum, this appears to be a major issue reported by many users for many of the Netgear NAS devices. There are all sorts of arguments as to why its so slow, and I do not understand why it should be, it makes no sense at all. This is a negative selling point for a netgear NAS: the Ultra 2 Plus is not a slow device, more so as it also applies to newer devices.
I have setup RAID1 arrays on PCs using motherboard hardware, add on RAID boards and using MS disk management, and all of these created the array in minutes ,that said they were using 100-500GB disks not 2TB discs, but faster processors. It seems we are going backwards!
I have found an article which suggests changing some parameters with SSH, I have never used SSH :
The article is not clear on the syntax of the commands {a..1} bit, exactly what should this be for 2 discs? Nor how to back it out afterwards.
The 2 x 2TB WD Green HDDs in the Ultra 2 are quite old, I decided not to replace as I was not getting any errors reported. These older discs may be contributing to the slow speed, if they have bad sectors, but the sync process is not reporting this - that is if would do anyway.
To try and speed it up, I have rebooted the Ultra 2, not sure it made much difference, it carried on where it left off.
Should I stop it and do a full "Disk Test" - how long will that take, will it make any difference and will it then start all over again?
Should I buy new disks for the Ultra 2? I dont want to spend best part of £250 for no good reason.
Any help appreciated, thanks.
6 Replies
2x2TB RAID-1 sync shouldn't take that long if you have healthy disks. I am wondering how long it took to reach the "68 hours to go" - you give the time estimates, but not the actual times.
OS 6 isn't as aggressive in reporting reallocated or pending sectors as OS 4.2 is. So I do recommend looking at the SMART stats - either by downloading the log zip file, and looking in disk_info.log, or by using ssh directly.
Another approach (better if you can do it) is to connect the two 7-year old drives to a windows PC, and test them with WDC's Lifeguard. After running the long read test, run the long version of the destructive write-zeros test. I've found the write-zeros test will find issues that the non-destructive tests miss.
lencw wrote:
I have setup RAID1 arrays on PCs using motherboard hardware, add on RAID boards and using MS disk management, and all of these created the array in minutes ,that said they were using 100-500GB disks not 2TB discs, but faster processors. It seems we are going backwards!
I don't think you are comparing apples with apples here. Your NAS has already created the array, what it is doing now is syncing it. With RAID-1, that amounts to reading all 2 TB from the first drive and writing it to the second.
A healthy WD20EARS delivered sequential read-write performance between 50 MB/s and 100 MB/s (depending on what part of the platters were being used). If we assume 75 MB/sec average, the disk I/O for a full sync would take about 7-8 hours. Note this is the best possible time - limited by the disks themselves, not the CPU or RAID hardware (or in the case of your NAS, the MDADM RAID software built into linux). Other disk I/O (for instance copying files to the volume while it is syncing) will slow this down substantially, due to the head seeking time.
It is possible to accelerate volume creation, by not bothering to sync at all (instead just mirroring as the sectors are written to the file system). It's possible that the MS system was doing that. However, MDADM doesn't.
FWIW, the WD20EFRX drives are faster than the old Greens (sequential speed ranges from about 60 MB/s to about 150 MB/s). Doing the same math, the best-case sync time for 4x2 TB RAID-10 would be about 11 hours. So 24 hours is long (and that was the issue what your linked-in thread was focused on). I believe that issue is specific to RAID-10. 4x4TB RAID-5 sync generally takes about 24 hours, and that requires twice the disk I/O of 4x2TB.
Thanks for your advice and comment, much appreciated.
Its now 64.1 hours to go, 39.4% complete. Its been running now for 44 hours. This is the Ultra 2 plus (on OS6) with the 2 green drives.
I am beginning to think its not worth continuing with the old 7year old green drives, as this upgraded NAS will be a backup for the new one. I probably should just get 2 new red drives. If I start hooking up to a PC and runnnig tests that could take ages, and I guess would not improve the current position.
The time for the new nas to sync the new RN214 FelxRaid R10 was probably an over estimate, I cant remember the exact time, definitely over 12 hours.
This older NAS is the 2 bay, so I will probably get 2 x 4TB WD red drives for it. Form what you say that should sync in well under 24 hours, have I understood correctly?
I am not clear about this, and never used MDADM. I am looking for the simple solution that will be reliable, I thought mirroring was syncing!
It is possible to accelerate volume creation, by not bothering to sync at all (instead just mirroring as the sectors are written to the file system). It's possible that the MS system was doing that. However, MDADM doesn't.
lencw wrote:
Its now 64.1 hours to go, 39.4% complete. Its been running now for 44 hours. This is the Ultra 2 plus (on OS6) with the 2 green drives.
I am beginning to think its not worth continuing with the old 7year old green drives, as this upgraded NAS will be a backup for the new one. I probably should just get 2 new red drives. If I start hooking up to a PC and runnnig tests that could take ages, and I guess would not improve the current position.
Aging drives is the most plausible explanation, though I still do have a 4x2TB array in my old NV+ that has these early disks, and it is still working. I use it as a secondary backup.
Lifeguard shouldn't take that long, less than one day to fully test a healthy 2 TB disk.
lencw wrote:
This older NAS is the 2 bay, so I will probably get 2 x 4TB WD red drives for it. Form what you say that should sync in well under 24 hours, have I understood correctly?
Correct. That's an estimate based on other posts here - I don't own the ultra plus, and don't have 4 TB drives installed. My pro-6 is 6x3TB though, and it will resync a new drive in less than 24 hours. My newer NAS use 6 TB and 8 TB drives.
lencw wrote:
I thought mirroring was syncing!
With RAID-1 it is. MDADM takes a purist approach - it creates a virtual disk from the physical disks in the array, and initializes the volume so the two disks exactly match. The quicker way doesn't match them in the beginning, taking advantage of the fact that the content of the unused free space doesn't matter.
Though later on it is helpful if everything matches - especially if you want to test the volume integrity.
NETGEAR Academy
Boost your skills with the Netgear Academy - Get trained, certified and stay ahead with the latest Netgear technology!
Join Us!
