I am currently working on a project where we are integrating Salesforce with SAP. Currently, we are trying to open our firewalls to allow traffic to/from cs18.salesforce.com, our Sandbox. However, we do get some errors when opening the IP's in our firewall. 

This is what we have opened:
96.43.146.188
96.43.146.189
96.43.146.60
96.43.146.61
96.43.150.188
96.43.150.189
96.43.150.60
96.43.150.61

Ports – 443, 1024–66535 (inclusive) 

This is the error we get:
Unable to create socket
com.sap.aii.af.ra.ms.api.RecoverableException: unable to create a socket: java.io.IOException: unable to create a socket'. Cause Exception: 'unable to create a socket' 

Any advice on where the issue may be? 

Hi,

 

I've got a batch job that works well in manual testing, processing multiple records in each batch without any issues. However my unit test fails if I set the scope greater than just 1 record with a "System.UnexpectedException: Error processing messages". I'm stumped, does anyone know what this cryptic exception means?

 

Thanks a lot,

Mike

I'm posting this in the hopes of helping "the next guy" that hits this.  

 

Our ant-based deployments, which had been going flawlessly for months, suddenly started failing, with lots of UNABLE_TO_LOCK_ROW errors in the logs from our various Apex test methods.  The same exact test methods passed when executed from "Run All Tests" in the UI.  We tried deploying with Eclipse and Change Sets, but got the same UNABLE_TO_LOCK_ROW errors.

 

The lines of code that were failing were all record inserts for certain objects.  We realized that if we removed those objects from our package.xml, we could complete the deployment without the errors.  However, we are making lots of object-level changes (new fields, etc), so this was a bad long-term solution.

 

After much troubleshooting, SFDC Tech Support finally found the root cause.  It turns out that if you modify the format of an auto-number field - say, change the name from "C-{0000}" to "CDM-{0000}" - subsequent deployment of that object will fail if any test methods insert records of that object.

 

Ouch.  Ugly, bad bug.  But once we knew the cause, we matched up all auto-number field formats in our source & destination orgs, and then re-ran the full ant script (with all objects) again, and the deployment was successful.

 

Hopefully SFDC will fix this odd bug quickly.  But in the meantime, if you get UNABLE_TO_LOCK_ROW" errors during deployment, check your auto-number field formats!