tree 906b0d7ba92875285f70ff59813d1694c7da0fff
parent 272a0aeb936724403e1f2fb6ccba0bd4548ae1f7
author Knut Olav Løite <koloite@gmail.com> 1634719844 +0200
committer GitHub <noreply@github.com> 1634719844 +1100
gpgsig -----BEGIN PGP SIGNATURE-----
 
 wsBcBAABCAAQBQJhb9hkCRBK7hj4Ov3rIwAAzPQIAFCl9tNbCTd/GXlkudQUf79g
 sLzm5hdCYR705jNUihYGDiH9HCbMq+QelY7wWF+zEos3oKhRhNp/ZjW5C82H6cTu
 x2zITc5ELAVrdFZjPt5miz0Xt1bQ7TeBK6KBFDGJuRUdPDA72j27bRBR8U/DjIaU
 UJiknAxPoQde1f4JrK3x6F0Uv/I0QY8s+3kByYn50pKi6p97vuMk0d+CkOSPm4o+
 4K0V1XT2FieQ9h9hjJjCm7YnJOHC1Hf7fgcjEYYOcFook78zOxQJFf/w0PdZRzey
 d0/MFAJ7pSUvHAFDGsxvwonFV4iMr1iKwdGPydvOiW11o0v5X1bZbi5TDMyLWls=
 =8z1K
 -----END PGP SIGNATURE-----
 

test(spanner): fix potential data race in test (#5006)

Fixes a potential data race condition in the test cases that use the TestExporter.
The TestExporter keeps all exported spans in a map that should only be accessed while the
mutex has been locked. The mutex is however not exported, which makes it impossible for tests
to lock it. The test cases therefore now only create one session, and wait for this session
to have been created, before it tries to access the exported spans. This prevents a potential
race condition where a BatchCreateSession RPC invocation tries to add an element to the map
with exported spans, while the test case is reading the map to check which spans have been
exported.

Fixes #5005