-
Hi,
I am trying to call the SOutput->resizeResBuf() function from the UDF. However, the slave traps giving a "symbol lookup error" for this function. Am i supposed to be uploading/linking this library to something else ? The UDF I am modifying is "wordbucket.so" supplied with the examples.
Thank you,
Saurabh.
2009-12-14 06:38:51 UTC in SECTOR
-
I ran into this same issue tonight. Basically, with version 1.24, I tried to test uploading of very small sizes - from 4 bytes to 8 bytes. The upload tool would hang and not respond. It should not take more than a minute to upload files of 4 bytes, even when I had my replication set to 2. I downgraded to 1.23 , and tried again with the small files, which were uploaded successfully.
I tried...
2009-12-07 07:42:47 UTC in SECTOR
-
Hello,
I was wondering, is there an API or tool available to get an approximate wait time for a job in queue to start ?
Thank you,
Saurabh.
2009-10-26 02:58:57 UTC in SECTOR
-
EDIT: By "Simple" files I mean, just running wordcount on 1 file of size no more than 200KB.
2009-10-13 06:29:08 UTC in SECTOR
-
Hello All,
I am having some problems running the "wordcount" DLL on Sphere.
There are a couple of questions I have:
1)
I had to directly copy all the files into the "data" directory of the slaves. How long does it typically take the metadata to refresh ? I would issue "ls" commands from the "tools" directory, but the files did not start...
2009-10-13 06:27:44 UTC in SECTOR
-
Hello,
Will it be possible to gain early access to the BETA code that supports running executables on Sphere instead of DLLs ?
I would really like to play around with this feature with SAGA and SAGA MapReduce in conjunction with Sphere MapReduce. Please let me know, if it will be possible to obtain this copy for running some experiments.
Thank you,
Saurabh.
2009-08-21 19:50:08 UTC in SECTOR
-
The point you mentioned is in regards to the number of files when compared to the number of slaves. When does the size of the individual files themselves matter ? i.e. if a single file is very large in a dataset, would it not become a performance bottleneck for the SPE assigned to process it ?.
2009-07-30 15:07:59 UTC in SECTOR
-
Thank you !
I posted one more question on the forum regarding Sector file sizes. It would be great if you can comment on that too ! If you can please give me some input as soon as it may be possible, I can work/ponder over this over the weekend. That would be very helpful.
Thank you,
Saurabh.
2009-07-30 15:01:36 UTC in SECTOR
-
Hi,
I had a question about how to calculate the optimal file sizes to upload to Sector. In your paper ("Sector and Sphere: A design and Implementation of a High Performance Data Cloud") , you mention, since Sector manages files, on the file level and NOT the block level:
"... it does require the user to break up large-datasets into multiple files or use a utility to...
2009-07-30 12:44:29 UTC in SECTOR
-
Hello Yunhong,
You are welcome !
I am not a POSIX expert :), but I *think* the file should exist if it is opened.
This, however, then leads to to other issues such as Read/Write locking
from applications that might concurrently access the file.
I am not sure if this is functionality you plan to support ? or maybe this is
already handled under the covers of Sector ?
Best Wishes,
2009-07-29 06:38:34 UTC in SECTOR
