CDS Issue during this last weekend (Saturday 15th - Sunday 16th of June)


Dear User,

Last weekend there was a CDS issue with one of the adaptors that provides access to some of the CDS catalogued data products. In particular this affected datasets hosted in MARS archive system at ECMWF. Such as ERA5 Complete, Seasonal forecast and UERRA.

The issue caused that requests run between Saturday 15th around 18h (BST) and 16th around 17h (BST) failed with an error message similar to "the request you have submitted is not valid" with detailed information containing references like the following:

mars - INFO - 20190616.041141 - Calling mars on 'marstunnel', callback on 52597
mars - INFO - 20190616.041141 - Callback at address, port 52597
mars - ERROR - 20190616.041326 - Failed to read 1
mars - ERROR - 20190616.041326 - Bad tag '0' (should be 'unsigned long long')
mars - ERROR - 20190616.041326 - Failed to read 4
mars - ERROR - 20190616.041326 - Failed to read 4
mars - WARN - 20190616.041326 - dhs_wait: stream_read_ulonglong: got error -32
mars - WARN - 20190616.041326 - Error -32 received while calling server on 'marstunnel', retrying.

The issue was solved Sunday 16th around 17h and access to the data was restarted since then. However, you must resubmit your failed requests to recover the lost requests. Taking advantage of this communication, CDS encourages you to rationalize the scope of your requests.

For more detailed information, please contact CDS user support.  

Thanks a lot and  Apologies for any inconvenience caused.


CDS Team