A temporary workaround to allow using the VMOD in vcl_backend_.* subroutines has been added (see https://github.com/varnish/libvmod-curl/issues/23). The VMOD behaves as usual, but you should always complete execution of each cURL request during a single VCL phase (i.e. vcl_recv, vcl_backed_response, vcl_deliver, etc.). Otherwise, some unexpected behavior may arise. Anyway, this is the usual approach when using the VMOD.
A temporary workaround to allow using the VMOD in
vcl_backend_.*
subroutines has been added (see https://github.com/varnish/libvmod-curl/issues/23). The VMOD behaves as usual, but you should always complete execution of each cURL request during a single VCL phase (i.e.vcl_recv
,vcl_backed_response
,vcl_deliver
, etc.). Otherwise, some unexpected behavior may arise. Anyway, this is the usual approach when using the VMOD.