Skip to content

[Bioc-devel] Time-out error on Ubuntu 20.04 build

3 messages · Selles Vidal, Lara, Hervé Pagès

#
Dear All,

Since it seems the main Linux target in 3.13 is now Ubuntu 20.04, I have been recently trying to get back to fixing the TIMEOUT of my package ncRNAtools.

Mike previously reported as seen below that he does not get the TIMEOUT when running R CMD build / check on an Ubuntu 20.04. I have also tried to reproduce the error on an Ubuntu 20.04, and could not manage to do so.

I am unsure what to do at this stage in order to trouble shoot the problem. Does anybody have any suggestions about to go around this?

Thanks a lot in advance,

Best wishes,

Lara



Lara Selles
lara.selles12 at imperial.ac.uk
#
Hi Lara,

I took a closer look at this and yes there seems to be something about 
nebbiolo1 that causes problems: it's behind a firewall. And for some 
reason this firewall causes the internal calls to httr::GET() to return 
responses with the following header:

   > GET(redirectURL)$headers
   $date
   [1] "Tue, 04 May 2021 01:59:02 GMT"

   $server
   [1] "Apache    adding: CentroidFold/ (stored 0%), 
CentroidFold/base-pairing-prob.txt (deflated 58%), 
CentroidFold/structure.eps (deflated 58%), 
CentroidFold/base-pairing-prob.png (deflated 18%), 
CentroidFold/base-pairing-prob.pdf (deflated 10%), 
CentroidFold/structure.pdf (deflated 19%), CentroidFold/structure.txt 
(deflated 25%), CentroidFold/base-pairing-prob.eps (deflated 60%), 
CentroidFold/structure.png (deflated 13%)"

   $`transfer-encoding`
   [1] "chunked"

   $`content-type`
   [1] "text/html; charset=UTF-8"

   $via
   [1] "1.1 phswsa8.partners.org:80 (Cisco-WSA/12.5.1-035)"

   $connection
   [1] "keep-alive"

   attr(,"class")
   [1] "insensitive" "list"

Two things about this header:

   1) The "via" field indicates the presence of the firewall.

   2) There's no "  adding" field. My understanding is that 
ncRNAtools:::checkSecondaryStructureQuery() expects to see that field as 
an indication that the initial query is no longer running. Note that the 
initial query here is no longer running and the final results are 
present in the header. However, instead of showing up in the "  adding" 
field, they've been merged to the "server" field!

For comparison here is what a clean header looks like on my laptop or on 
other build machines:

   > GET(redirectURL)$headers
   $date
   [1] "Tue, 04 May 2021 02:13:18 GMT"

   $server
   [1] "Apache"

   $`  adding`
   [1] "CentroidFold/ (stored 0%), CentroidFold/base-pairing-prob.txt 
(deflated 58%), CentroidFold/structure.eps (deflated 58%), 
CentroidFold/base-pairing-prob.png (deflated 18%), 
CentroidFold/base-pairing-prob.pdf (deflated 10%), 
CentroidFold/structure.pdf (deflated 19%), CentroidFold/structure.txt 
(deflated 25%), CentroidFold/base-pairing-prob.eps (deflated 60%), 
CentroidFold/structure.png (deflated 13%)"

   $connection
   [1] "close"

   $`transfer-encoding`
   [1] "chunked"

   $`content-type`
   [1] "text/html; charset=UTF-8"

   attr(,"class")
   [1] "insensitive" "list"

Unfortunately I don't have any control on how the firewall at DFCI is 
configured so I'm asking if you would be willing to modify ncRNAtools to 
accommodate this situation. Note that the benefit of doing this is that 
it will also solve the issue for people in other institutions that have 
a firewall configured in a way similar to the DFCI's firewall.

Hope this makes sense.


Best regards,
H.
On 03/05/2021 14:37, Selles Vidal, Lara wrote:

  
    
1 day later
#
Hi Herv?,

Thanks a lot for looking with detail into this. I completely agree that a fix for this would be very beneficial.

I pushed yesterday an update to the devel version of ncRNAtools, which hopefully should fix the issue.

The change seems not to be propagated yet, but hopefully it will be reflected in the build report soon!

Best wishes,

Lara


Lara Selles
lara.selles12 at imperial.ac.uk