[basickler]

The highest cluster density I've seen (basically top of the PF curve) is around 311,000 PF clusters/tile at 375,000 RAW clusters/tile. On a GAIIx with 120 tiles in a 2x100 run this would give you close to 60G a flowcell. And yes I think this is limitation based on a combination of software and resolution. Push the density much farther and you'll actually start to see PF drop (this has been the case for every version of the pipeline).

Btw, one of the main limiting factors in the hard drive I/O is the awful performance of Dell's PERC5/6 Raid10.

[olus]

Charlie,

We have experienced exactly those problems with delayed transfer & paused runs on our latest run. I contacted Illumina and they tried to pass it off as problems with our network. The instrument we are having trouble with was originally delivered as a 1G Genome Analyzer (i.e. the original model) but has been upgraded twice now to make it a GA IIx. However it still has the older Dell 690 workstation with a 930 GB D: drive array.

Hello, I just want to share with you.
We'have the same hardware configuration here and the same issue (beginning from SCS2.6 upgrade).
We saw a big delay around cycles 4/5 until 8/9, then normalizes with a second 'peak' around the 17th cycle. The delays are in the order of 5 ours per cycle (normally is less than 1h).
It doesn't seem related to cluster density, neither on the disk space, infact we're not saving images at all. The Dell and the linux server are connected by a point-to-point cable so there is no way to have delays due to lan traffic.
Our FAS has no idea on what is going on, we run 3 cells so far with this 'delays' , but we had no other problems related to this.
Let's follow up.

Cheers

[bpinsea]

Just to share our experiences (and FAS input), With RTA1.6 we have been seeing similar issues, and it is particularly bad on the machines with older/smaller hard drives. Even if you aren't saving images, the software appears to wait and process data in 5 cycle batches, so you will get at least 5 cycles worth of image data on the drive at any given time. Also, once the chemistry completes, the software downgrades the importance of deleting the images after processing. This has been confirmed by our FAS. If your processing is too far behind your chemistry, any images on the drive for processing will remain there until the software completes - our drives started filling at a rate of >30GB an hour without any additional image data being added. The data is salvageable, but it makes for a rough few days.

[basickler]

Removed comment.

[cnicolet]

problems have gotten better

Hi all:
Thanks for everyone's comments and similar war stories. We changed a couple of things, not sure if they were the problem but runs have gotten more consistent with minimal or no pausing. First we disabled the recycle bin on the D drive, as suggested by basickler (and I was told this was buried in a Tech Note too--oops!). We also were getting Windows system time sync error messages because our Dells were trying to sync to an internet time server that was unavailable. We now have them time sync to the same server images are sent to (the server can access the internet), and that seems to have helped. Good luck everyone!

[BerkeleyGSL]

SCS2.6, RTA1.6 update available

I noticed Illumina slipped an update to SCS2.6 and RTA1.6 on iCom the other day. Mostly updated quality files, and some bug fixes to RTA. Didn't read the notes carefully to figure out what they fixed. OLB1.6 was also updated to 1.6.1 with the new quality tables. Don't know if this will fix any of the issues from this thread.